AI 行业雷达
中EN

全部 AI 动态

1384 条正在获取最新内容

按来源、主题或关键词浏览完整信息流。

来源
分类

最新动态

9月29日周二

9月29日周二

llama.cpp1 个来源

b11242

<details open> mtmd: fix GCC 15 stringop-overflow in decode_embd_batch (#29607) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50949667> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **L

其他1 个来源
来源与时间线
llama.cppb11242
llama.cpp1 个来源

b11240

<details open> server : support typed content (vision/audio/video) input for /v1/embeddings endpoint (#29556) * server : support multimodal input for /v1/embeddings (Qwen3-VL-Embedding) Accept the OpenAI-style wrapped content array format for multimodal embedding requests. Each {"content": [...]} object is one input that produces one embedding; text parts are concatenated and image_url parts are decoded via handle_media then spliced with process_mtmd_prompt. The legacy formats (plain string, token arrays, mixed arrays, and the {prompt_string, multimodal_data} object) continue to work unchanged via tokenize_input_prompts. Bare content arrays (the unwrapped shape) are rejected with a migration

其他1 个来源
来源与时间线
llama.cppb11240
llama.cpp1 个来源

b11239

<details open> vulkan: include functional header (#29597) Fixes the compile error: `no template named 'function' in namespace 'std'` </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50927029> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/relea

其他1 个来源
来源与时间线
llama.cppb11239
The Verge AI1 个来源

AMD is acquiring AI company World Labs in a deal worth more than $8 billion

AMD announced today that it's acquiring World Labs, an AI research lab co-founded by the prominent researcher Dr. Fei-Fei Li, in an all-stock deal worth approximately $8.2 billion. World Labs launched in 2024 and was valued at $1 billion in a matter of months. The startup launched its first commercial product, a world generation model […]

其他1 个来源
来源与时间线
The Verge AIAMD is acquiring AI company World Labs in a deal worth more than $8 billion
Ollama1 个来源

v0.35.0

Decision models Ollama now supports decision models through `/v1/systemone`, based on TypeSafe’s Jev API. Decision models return choices, probabilities, and scores instead of text. Use them for tasks such as ticket triage, model routing, and content classification. Available models: - **Nimble** from Bespoke Labs - **Tev1** from Together AI ```sh ollama pull nimble ``` Send context and one or more questions: ```sh curl http://localhost:11434/v1/systemone \ -H 'Content-Type: application/json' \ -d '{ "model": "nimble", "state": "Our checkout has returned 500 errors since 9am.", "questions": { "la

其他1 个来源
来源与时间线
Ollamav0.35.0
Ollama1 个来源

v0.35.0

Decision models Ollama now supports decision models through `/v1/systemone`, based on TypeSafe’s Jev API. Decision models return choices, probabilities, and scores instead of text. Use them for tasks such as ticket triage, model routing, and content classification. Available models: - **Nimble** from Bespoke Labs - **Tev1** from Together AI ```sh ollama pull nimble ``` Send context and one or more questions: ```sh curl http://localhost:11434/v1/systemone \ -H 'Content-Type: application/json' \ -d '{ "model": "nimble", "state": "Our checkout has returned 500 errors since 9am.", "questions": { "la

其他1 个来源
来源与时间线
Ollamav0.35.0
The Verge AI1 个来源

OpenAI’s AI agents need to catch up

OpenAI popularized the modern generative AI chatbot, but as its 2026 DevDay event approaches, it's fallen behind in one of the industry's hottest categories: continuously running, consumer-facing AI agents. On Tuesday, it will likely try to capture the lead in that race. Rumors abound that OpenAI will release its own AI agent, dubbed Aeon - […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s AI agents need to catch up
The Verge AI1 个来源

AI is supercharging hacking, and your local hospitals and banks aren’t ready

In March, Janice Malone began getting calls about suspicious activity from her nonprofit organization, Vivian's Door. Vivian's Door, headquartered in Alabama, typically provided training, resources, and community to underserved and minority-owned businesses. The work sometimes put it in close contact with these companies' financial data, which was stored on its systems. But suddenly, concerned callers […]

其他1 个来源
来源与时间线
The Verge AIAI is supercharging hacking, and your local hospitals and banks aren’t ready
Anthropic Python SDK1 个来源

v1.9.0

1.9.0 (2026-09-28) Full Changelog: v1.8.0...v1.9.0 Features * **api:** add between_tools thinking type (a9a577d) * **api:** add claude-sonnet-5-5 (a5a25e9) * **api:** add include_inherited and source to workspace rate limits (0bf4af8) * **api:** add typed event type values to the Managed Agents events list f

其他1 个来源
来源与时间线
Anthropic Python SDKv1.9.0
MIT Technology Review AI1 个来源

When can we say AI made a scientific discovery?

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Last Wednesday, Anthropic announced that earlier this year it had launched a molecular biology lab, where Claude agents read and conjecture about hard biology problems and human scientists run experiments on what…

其他1 个来源
来源与时间线
MIT Technology Review AIWhen can we say AI made a scientific discovery?
OpenAI Python SDK1 个来源

v3.20.0

3.20.0 (2026-09-28) Features * **api:** add Agents credential and session options (#3967) (bb68198) * **api:** add Cyber access programs to Responses (#3956) (09c5b6f) * **responses:** opt in to incremental WebSocket text and tool snapshots (#3973) ([d0207b4](https://github.co

监管1 个来源
来源与时间线
OpenAI Python SDKv3.20.0
Google Gemini Blog1 个来源

See what 4 builders are making with Gemini 3.8 Flash

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Builders_Gemini_3.8_Flash.max-600x600.format-webp.webp">See what developers have built with Gemini 3.8 Flash, our most intelligent workhorse model, launched in early September.

其他1 个来源
来源与时间线
Google Gemini BlogSee what 4 builders are making with Gemini 3.8 FlashGoogle Gemini Blog3 ways this grocer cooks for 200 guests with Gemini

9月28日周一

MIT Technology Review AI1 个来源

Who’s liable when AI agents go rogue?

MIT Technology Review Explains: Let our writers untangle the complex, messy world of technology to help you understand what’s coming next. You can read more from the series here. Over the past few months, a cascade of cyberattacks by AI agents has stunned the world. In July, OpenAI disclosed that a swarm of its agents…

其他1 个来源
来源与时间线
MIT Technology Review AIWho’s liable when AI agents go rogue?
llama.cpp1 个来源

b11223

<details open> server : allow RANK pooling batch splitting for causal LLM rerankers (ie. Qwen3 and Qwen3-VL) (#28876) * server : allow splitting RANK pooling for causal LLM rerankers Rerank models fall into two categories: bidirectional cross-encoders (BERT, etc.) that require all tokens in a single physical batch, and causal LLMs repurposed as rerankers (Qwen3, Qwen3-VL) that can use chunked prefill like any other decoder. Previously the server rejected all RANK-pooling inputs larger than n_ubatch, and the graph builder hardcoded QWEN3/QWEN3VL arch checks to determine last-token pooling. This broke long-document and multimodal reranking for causal models. Fix: expose llama_get_causal_attn(c

其他1 个来源
来源与时间线
llama.cppb11223
The Verge AI1 个来源

Engram is a sampler that turns broken AI hallucinations into music

Music startup Thoughtful Things has just launched the Kickstarter campaign for its first instrument, Engram. It's a sampler and groovebox that uses AI to mangle incoming audio and even hallucinate completely new sounds. This isn't Suno in a box, though. This isn't a "push-button, get-song" device, aimed at creating something that sounds ready for top-40 […]

其他1 个来源
来源与时间线
The Verge AIEngram is a sampler that turns broken AI hallucinations into music
llama.cpp1 个来源

b11222

<details open> common : avoid side effects around params parsing (#29537) - register --rpc unconditionally and call llama_supports_rpc() only from its handler - print server "initialization ..." log after args are parsed Assisted-by: pi:llama.cpp/MiMo-V2.6-Flash-RL </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50580981> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/

其他1 个来源
来源与时间线
llama.cppb11222
The Verge AI1 个来源

OpenAI agents tried to ‘bruteforce’ a UN website

Security researcher Rowan Howard-Jones says that OpenAI agents scanned the UN Conference on Trade and Development's (UNCTAD) statistics site over 16,000 times between April and June. While the incident doesn't quite rise to the level of the Hugging Face hack, or the recent attacks on US government sites, it's yet another concerning example of AI […]

其他1 个来源
来源与时间线
The Verge AIOpenAI agents tried to ‘bruteforce’ a UN website
llama.cpp1 个来源

b11221

<details open> common : make string_split<T> throw on invalid values (#29518) Signed-off-by: Adrien Gallouët <angt@huggingface.co> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50573055> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/release

其他1 个来源
来源与时间线
llama.cppb11221

9月27日周日

llama.cpp1 个来源

b11205

<details open> cuda: support Nemotron 3 Puzzle state size 96 for ssm scan (#28717) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50461180> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework

其他1 个来源
来源与时间线
llama.cppb11205
llama.cpp1 个来源

b11203

<details open> cuda: add F16 input to the FWHT (#29096) * cuda: add F16 input to the FWHT The CUDA FWHT accepts F32 input only. This makes the source type a template parameter, so the kernel reads an F16 source directly instead of requiring a converted copy. The F32 path is unchanged. supports_op accepts an F16 src1 against an F32 src0 for the Hadamard hint. Every other F16 src1 against a non-F16 src0 is still refused. ggml_cuda_op_mul_mat_use_fwht is the single predicate both supports_op and the dispatch call now share, checking contiguity and same-shape(src1, dst) in addition to the type/hint conditions above. Without a shared predicate, supports_op could admit an op that ggml_cuda_op_fwht

其他1 个来源
来源与时间线
llama.cppb11203
llama.cpp1 个来源

b11202

<details open> server : fix wake_fd warning on Windows (#29479) Signed-off-by: Adrien Gallouët <angt@huggingface.co> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50455005> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11

其他1 个来源
来源与时间线
llama.cppb11202
The Verge AI1 个来源

OpenAI pauses training of its ‘most capable models’

As reports of OpenAI's models breaking containment, hacking sites, and generally getting out of control pile up, the company has made the decision to pause training of its most powerful models. The decision was made after a model being tested within a sandbox exploited a loophole to gain internet access. The incident happened on September […]

安全1 个来源
来源与时间线
The Verge AIOpenAI pauses training of its ‘most capable models’

9月26日周六

The Verge AI1 个来源

Can Cloudflare CEO Matthew Prince save the web from AI?

Today, I’m talking with Matthew Prince, who is CEO of Cloudflare. This episode is part of a two-part series on the future of business. Matthew last joined us on the show about two and a half years ago, at what we thought then was a wild pivot point for the internet — and now it […]

其他1 个来源
来源与时间线
The Verge AICan Cloudflare CEO Matthew Prince save the web from AI?
llama.cpp1 个来源

b11192

<details open> vendor : update cpp-httplib to 0.58.0 (#29407) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50337762> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu

其他1 个来源
来源与时间线
llama.cppb11192
llama.cpp1 个来源

b11191

<details open> common,rpc : simplify fs_create_directory_with_parents() (#29432) The original function was broken on Windows for some unicode paths Paths without a trailing separator now create the last directory too, matching the function name. All current callers already include a trailing separator, so this change does not affect them. Signed-off-by: Adrien Gallouët <angt@huggingface.co> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50291720> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, K

其他1 个来源
来源与时间线
llama.cppb11191
llama.cpp1 个来源

b11190

<details open> mtmd: fix mel preprocessor in LFM2 audio (#29403) which resulted in different greedy transcripts for 4.5% of English and 6.5% of Japanese test utterances. In Japanese, some differences changed entire words. This change: * uses `log(x + 2^-24)` instead of clamping to the log floor * uses a symmetric Hann window, equivalent to `torch.hann_window(periodic=False)` * adds the normalization epsilon to the standard deviation instead of inside the square root Only the `lfm2a` preprocessor opts into these behaviors. Other audio preprocessors are unchanged. Tested on top of 84e76d8 using `llama-server` with CUDA and `temperature=0`, compared against http://github.com/Liquid4All/liquid-a

其他1 个来源
来源与时间线
llama.cppb11190
The Verge AI1 个来源

Meta makes the Muse filesystem even more accessible

Yesterday, with a little prodding, it was discovered that Meta's Muse would expose its filesystem to curious users. The files offered a fascinating peek under the hood of an AI chatbot, and appeared to expose details we weren't meant to see, not least because Muse itself told people, including us, it wasn't supposed to reveal […]

其他1 个来源
来源与时间线
The Verge AIMeta makes the Muse filesystem even more accessible

9月25日周五

The Verge AI1 个来源

Sony and UMG are suing Suno again

Sony and Universal Music Group filed yet another suit against Suno. The labels claim its new v6 model still infringes on their copyrights because it's trained on user outputs from previous models, which were themselves trained on unlicensed music ripped from YouTube and other sources. Sony and UMG are notable holdouts who did not sign […]

监管1 个来源
来源与时间线
The Verge AISony and UMG are suing Suno again
The Verge AI1 个来源

One company is at the center of a wave of rogue AI attacks

In July, OpenAI revealed that its AI agents had attacked Hugging Face without permission, sparking widespread concerns about AI safety. Since then, a string of similar incidents involving agents from Meta, Anthropic, Google, and other companies has fueled further fears about rogue AI. As disclosures implicating numerous AI models trickled out over the past few […]

其他1 个来源
来源与时间线
The Verge AIOne company is at the center of a wave of rogue AI attacks
MIT Technology Review AI1 个来源

The Pentagon wants $30 million to build an AI-powered lie detector

The US government wants to spend $30.3 million over the next five years on an improved form of lie detector, according to a Department of Defense budget request. The program, called Polygraph+ or Polygraph Next, will focus on scoring algorithms that use artificial intelligence and machine learning and on a technique called “standoff sensing,” which…

其他1 个来源
来源与时间线
MIT Technology Review AIThe Pentagon wants $30 million to build an AI-powered lie detector
Ollama1 个来源

v0.40.0

What's Changed **Models run on MLX on Apple Silicon by default** In this release, on Apple Silicon devices, model architectures supported by the MLX runtime will automatically run on MLX. ``` ollama pull qwen3.8 ollama run qwen3.8 ``` During the pre-release we will be testing and enabling additional models. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.4...v0.40.0-rc0

其他1 个来源
来源与时间线
Ollamav0.40.0
llama.cpp1 个来源

b11173

<details open> metal : fix graph capture and handle empty graphs (#29390) - return early when the graph has no nodes - drop the redundant reset of capture_compute: the decrement at the top of the function already transitions the counter from 0 to -1, so a capture happens exactly once - hint at METAL_CAPTURE_ENABLED=1 in the capture error message - pass capture_compute == 0 (not the raw counter) as use_capture to ggml_metal_op_init, so GPU debug-group markers are only emitted on the captured compute Assisted-by: pi:llama.cpp/Qwen3.8-27B </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50032954> **macOS/iOS:** - [macOS Apple

其他1 个来源
来源与时间线
llama.cppb11173
llama.cpp1 个来源

b11172

<details open> metal : optimize sparse FA + clean-up (#29377) * metal : cache sparse FA indices in shared memory Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * metal : simplify shared memory size calculation Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * pi : update general * metal : unroll sparse index load Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50026590> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI e

其他1 个来源
来源与时间线
llama.cppb11172
llama.cpp1 个来源

b11171

<details open> sync : ggml (#29396) * ggml : bump version to 0.25.2 (ggml/1642) * ggml : fix ubsan error in `ggml_graph_nbytes` (ggml/1644) * ggml : bump version to 0.25.3 (ggml/1645) * sync : ggml </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/50020953> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11171/llama-b11171-bin-macos-x64.tar.gz

其他1 个来源
来源与时间线
llama.cppb11171
The Verge AI1 个来源

Gemini 3.8 Live with Live Avatar gives Google’s AI a face

Google's new Gemini 3.8 Live update lets users have conversations with the model while watching an animated AI persona respond in real time. The "Live Avatar" will lip-sync and show different facial expressions during conversations, but it's currently only available to Gemini Enterprise customers. As noted by Google, Live Avatar can transition between the 97 […]

其他1 个来源
来源与时间线
The Verge AIGemini 3.8 Live with Live Avatar gives Google’s AI a face
The Verge AI1 个来源

Jensen Huang talks about AI and climate change like a supervillain

As Jensen Huang puts it, AI can help fight climate change - but only if it inflicts "an enormous amount of pain and suffering" first. The Nvidia CEO discussed the future of energy and AI's impact on our planet in the latest episode of The Ezra Klein Show. But his comments boil down to the […]

其他1 个来源
来源与时间线
The Verge AIJensen Huang talks about AI and climate change like a supervillain
The Verge AI1 个来源

Meta is going to let you build games with AI right on your phone

Meta has a new plan to get people to make games for its Horizon social platform. The company today announced two new development tools that will let you create games with AI prompts: Horizon Create, a mobile app, and Horizon Studio, a browser app that offers more granular controls. The apps will be available in […]

其他1 个来源
来源与时间线
The Verge AIMeta is going to let you build games with AI right on your phoneThe Verge AIMeta is making a standalone Muse AI gadgetThe Verge AIMuse is coming to Meta smart glassesThe Verge AIMeta ditches the camera on its newest smart glasses
Google Gemini Blog · Google DeepMind Blog2 个来源

Introducing Gemini 3.8 Live with Live Avatar

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Slide_16_9_-_37.max-600x600.format-webp.webp">Introducing Gemini 3.8 Live with Live Avatar, which brings near real-time visual presence to Gemini’s conversational AI.

其他2 个来源
来源与时间线
Google Gemini BlogIntroducing Gemini 3.8 Live with Live AvatarGoogle DeepMind BlogIntroducing Gemini 3.8 Live with Live AvatarGoogle DeepMind BlogIntroducing Gemini 3.8 Live and 3.8 Live Extended ThinkingGoogle DeepMind BlogIntroducing Gemini 3.8 Flash and 3.8 Flash CyberGoogle DeepMind BlogIntroducing agentic video understanding with GeminiGoogle Gemini BlogIntroducing Gemini 3.8 Live and 3.8 Live Extended Thinking

9月24日周四

OpenAI Python SDK1 个来源

v3.19.2

3.19.2 (2026-09-23) Bug Fixes * preserve single files for fallback extraction paths (#3875) (bfd3680) Chores * **api:** clarify approximate web search location defaults (#3953) (a95c95e) * **api:** clarify Realtime modality array definitions (#3954) ([e79cf53](https://gith

监管1 个来源
来源与时间线
OpenAI Python SDKv3.19.2
llama.cpp1 个来源

b11149

<details open> tests: add `-b/--backend` option to test-llama-archs for testing a specific backend (#27372) * tests: add backend option to test-llama-archs * Update tests/test-llama-archs.cpp Co-authored-by: Johannes Gäßler <johannesg@5d6.de> * remove extra space --------- Co-authored-by: Johannes Gäßler <johannesg@5d6.de> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49661383> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pul

其他1 个来源
来源与时间线
llama.cppb11149
llama.cpp1 个来源

v0.5.0

Overview This release focuses on backend performance and correctness, broader model coverage, and more robust server/router operation. It adds HRM-Text (DFM Mimir 1B) support, MiMo-V2.6 and HunyuanOCR conversion support, ggml 0.25.0 backend improvements, multi-address HTTP binding, image outputs from function calls, and several chat parser/UI fixes. Highlights - Accelerate CUDA `conv2d` with implicit GEMM (#29135) - Add Metal MoE and SSM_CONV fusion optimizations (#28948) - Allow the server to bind to multiple addresses (#28690) ##

其他1 个来源
来源与时间线
llama.cppv0.5.0
OpenAI Python SDK1 个来源

v3.19.1

3.19.1 (2026-09-23) Bug Fixes * **chat:** preserve single-pass tool iterables (#3770) (33ffa1f) * **client:** merge HTTP headers case-insensitively (#3486) (5e39766) Chores * **api:** clarify Chat Completions seed limits (#3945) ([be9d666](https://github.com/openai/openai-

监管1 个来源
来源与时间线
OpenAI Python SDKv3.19.1
llama.cpp1 个来源

b11147

<details open> opencl: add A8 Q6_K non-MoE dp4a binary kernel (#29057) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49630180> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** -

其他1 个来源
来源与时间线
llama.cppb11147

9月23日周三

MIT Technology Review AI1 个来源

The AI Hype Index: AI loves cheating

Brace yourself: It turns out AI is being optimized for cheating. OpenAI’s agents hacked into Hugging Face to get the answers to a cybersecurity test. Next, they solved a prestigious math problem (or just stole from two top mathematicians’ answer sheets). Anthropic’s models have also hacked into other companies’ systems four times already. And that’s…

其他1 个来源
来源与时间线
MIT Technology Review AIThe AI Hype Index: AI loves cheating
Ollama1 个来源

v0.34.4

What's Changed * server: fix intermittent "model not found" errors. by @rick-github in https://github.com/ollama/ollama/pull/18438 * server: apply structured outputs in a single pass on thinking models by @jessegross in https://github.com/ollama/ollama/pull/18479 * app: avoid System Events for ChatGPT/Codex detection by @hoyyeva in https://github.com/ollama/ollama/pull/18601 * llama.cpp: version update by @dhiltgen in https://github.com/ollama/ollama/pull/18577 * MLX: version bump by @dhiltgen in https://github.com/ollama/ollama/pull/18576 * mlx: select Gemma 4 image resolution dynamically by @dhiltgen in https://github.com/ollama/ollama/pull/18603 * mlx: speed up Qwen 3.8 prompt processi

其他1 个来源
来源与时间线
Ollamav0.34.4
Ollama1 个来源

v0.34.4

What's Changed - Structured outputs on thinking models now apply in a single pass, making them faster and more reliable. - Fixed intermittent "model not found" errors with a large local library - Fixed the macOS app becoming unresponsive when checking if ChatGPT or Codex is running. - Qwen 3.8 prompt processing is faster on Apple Silicon. - Gemma 4 on Apple Silicon now picks the best image resolution per image, keeping more detail in high-resolution images. - Updated llama.cpp, MLX, and XGrammar. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.3...v0.34.4

其他1 个来源
来源与时间线
Ollamav0.34.4
The Verge AI1 个来源

OpenAI wants to consult elite mathematicians about how to not fumble again

After turning a string of spectacular mathematical results into a reputational crisis, OpenAI is consulting human mathematicians to help it figure out a less disastrous path forward. On Monday, the company announced a new independent panel of mathematicians tasked with advising it and other AI companies on their interactions with mathematical research and the wider […]

其他1 个来源
来源与时间线
The Verge AIOpenAI wants to consult elite mathematicians about how to not fumble again
OpenAI Python SDK1 个来源

v3.19.0

3.19.0 (2026-09-22) Features * **api:** add GCP external storage support (#3943) (d12d60f) * **api:** add GPT-Rosalind research model (#3940) (d041d73) Bug Fixes * **_utils/_transform:** propagate __api_exclude__ in _async_transform_recursive (#3324) ([161ae65](https://git

监管1 个来源
来源与时间线
OpenAI Python SDKv3.19.0
TechCrunch AI1 个来源

TechCrunch Founder Summit’s agenda revealed: Unlock fundraising, hiring, and AI insights in Boston on November 4

Founders shouldn't have to learn the hardest lessons the hardest way. TechCrunch Founder Summit is designed to make the challenges of starting a company easier and the highs that much greater.

其他1 个来源
来源与时间线
TechCrunch AITechCrunch Founder Summit’s agenda revealed: Unlock fundraising, hiring, and AI insights in Boston on November 4
llama.cpp1 个来源

b11115

<details open> opencl: add bin kernel `kernel_gemm_noshuffle_q4_k_q8_1_dp4a_ila_a8_bin` (#29056) * opencl: add A8 Q4_K non-MoE dp4a binary kernel * opencl: rename binary kernel selection helpers </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49361383> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) -

其他1 个来源
来源与时间线
llama.cppb11115
llama.cpp1 个来源

b11114

<details open> server: fix router eviction races with the existing queue (#29217) * server: route every model load through the queue A model loaded by the fast path has no queue entry, so tick() evicts it at its LOADED transition before its own request is proxied. Every load now joins the queue, whose entry protects the model until its waiters leave. * server: do not admit requests into a stopping model A request for a model that is being stopped still sees it LOADED and is proxied into the dying child. Such a request now joins the queue and is served by the next instance. The stopping mark is cleared under the same lock that sets UNLOADED, so no request can see a model that is neither stopp

其他1 个来源
来源与时间线
llama.cppb11114
The Verge AI1 个来源

Rabbit’s new AI agent doesn’t need an R1 to run

Rabbit, the company behind the underwhelming R1 device, is rolling out a standalone AI agent that you don't need its hardware to use, as reported earlier by Wired. The startup says its new OS3 "agentic operating system" runs in the cloud but operates locally across Windows, Mac, and Linux devices. According to Rabbit, you can […]

其他1 个来源
来源与时间线
The Verge AIRabbit’s new AI agent doesn’t need an R1 to run
llama.cpp1 个来源

b11113

<details open> server: do not pass log file to children (#29212) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49344105> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubun

其他1 个来源
来源与时间线
llama.cppb11113
NVIDIA Generative AI Technical Blog1 个来源

Enabling Private High-Performance Production AI Inference with NVIDIA Confidential Computing

As large language model (LLM) inference increasingly processes sensitive information and proprietary model context across personal, enterprise, and regulated...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogEnabling Private High-Performance Production AI Inference with NVIDIA Confidential Computing
Microsoft Foundry Blog1 个来源

Claude Opus 5.5 comes to Microsoft Foundry for long-running coding and knowledge work | Microsoft Community Hub

AI models are increasingly taking on work that extends far beyond a single prompt: building a feature across a codebase, investigating a complex issue,...

监管1 个来源
来源与时间线
Microsoft Foundry BlogClaude Opus 5.5 comes to Microsoft Foundry for long-running coding and knowledge work | Microsoft Community Hub
The Verge AI1 个来源

Andreessen Horowitz is launching an ‘academy’ with no homework and partnerships with Palantir, Google, and Meta

Venture capital firm Andreessen Horowitz (a16z) is creating an "academy" positioned as a pipeline for young people to build or join a Silicon Valley startup. The "Horowitz Andreessen Academy" will launch with 10 partners, including Anduril, Anthropic, Coinbase, Google, Meta, Nvidia, OpenAI, Palantir, Replit, and Stripe, along with $42 million in funding led by a16z. […]

其他1 个来源
来源与时间线
The Verge AIAndreessen Horowitz is launching an ‘academy’ with no homework and partnerships with Palantir, Google, and Meta
Anthropic Python SDK1 个来源

v1.8.0

1.8.0 (2026-09-22) Full Changelog: v1.7.0...v1.8.0 Features * **api:** add support for claude-opus-5-5, inline tool definitions and MCP tool-list pinning (beta) (b5cc700) Bug Fixes * **api:** share one evaluated_permission enum across Managed Agents events (f4f51c8) * **streaming:** avoid a Python 3.13 crash at exit when a stream is left open ([af029bd](https://github.com/anthropics/anthropic-sdk-python/

其他1 个来源
来源与时间线
Anthropic Python SDKv1.8.0

9月22日周二

MIT Technology Review AI1 个来源

Roundtables: The Deadly Failures of The Virtual Border Wall

The US has spent billions building a “virtual wall” of surveillance towers along its southern border over the past 25 years, promising they will help detect and apprehend border crossers and save lives. But a groundbreaking investigation by MIT Technology Review has documented over a thousand people who moved through areas watched by these towers…

其他1 个来源
来源与时间线
MIT Technology Review AIRoundtables: The Deadly Failures of The Virtual Border Wall
MIT Technology Review AI1 个来源

Don’t be fooled by this summer of AI hype

It’s been a busy few months for AI hype. At the end of April, Anthropic claimed that its model Claude Mythos is better at finding software vulnerabilities than most security experts. Then we had the OpenAI–Hugging Face hacking incident, after which Anthropic (proudly) and Meta (reluctantly) disclosed similar incidents involving their models. This was followed…

安全1 个来源
来源与时间线
MIT Technology Review AIDon’t be fooled by this summer of AI hype
vLLM1 个来源

v0.30.0

v0.30.0 Highlights This release features 762 commits from 315 contributors (104 new)! * **New models**: DeepSeek-V4.1-Flash (#56214, #56228, #56208) with the whole KV stored in MXFP8 through the FlashMLA V4.1 record on SM100 (#56893), DeepGEMM Mega-mHC (#56962), and async Engram prefetch with Engram DP sharding (#56512); DeepSeek-V4-Flash-Vision-Exp (#54566), also on ROCm (#55107) and with LoRA (#55897); GLM-5.3-Flash (#53906) with EPLB (#55119); K2-Horizon (#55063); Cohere Compass (#54774); Bailing V3 VL (#55921); Nanbeige4.2 via the Transformers backend (#56071); and a DeepSeek-V4 CPU backend with AVX512/AMX sparse MLA, indexer, mHC and compressor kernels (#55355). * **Fast Start**: a

其他1 个来源
来源与时间线
vLLMv0.30.0
OpenAI Python SDK1 个来源

v3.17.0

3.17.0 (2026-09-22) Features * **api:** add external storage configuration management (#3909) (6332577) * **api:** add safety case retrieval (#3911) (a87b938) * **api:** add safety warning and deactivation webhook events (#3908) ([19f1f37](https://github.com/openai/openai-pyth

监管1 个来源
来源与时间线
OpenAI Python SDKv3.17.0
llama.cpp1 个来源

b11093

<details open> metal : fix mask bounds in flash attention block pre-pass (#29220) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49087452> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework *

其他1 个来源
来源与时间线
llama.cppb11093
llama.cpp1 个来源

b11090

<details open> cuda: fix sm_70 tile compilation error (#29224) The 5-argument load_ldmatrix added in 1884824fd only defines tile<16,8>, so the Volta tile<8,4> does not match. See https://github.com/ggml-org/llama.cpp/issues/29222 for details. Building on 1884824fd, generalize the tile shape of the 5-argument load_ldmatrix from <16,8> to <I,J>, so the non-swizzle branch forwards to the 3-argument loader for any shape. Local compilation and testing passed. Assisted-by: DeepSeek V4.1 Flash (OpenCode) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/49078119> **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggm

监管1 个来源
来源与时间线
llama.cppb11090
NVIDIA Generative AI Technical Blog1 个来源

Simplifying Model Serving Across Multiple GPUs with NVIDIA TensorRT Multi-Device Integration in NVIDIA Dynamo-Triton

The compute and memory demands of generative AI increasingly exceed what a single GPU can provide. NVIDIA TensorRT multi-device inference is a new capability...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogSimplifying Model Serving Across Multiple GPUs with NVIDIA TensorRT Multi-Device Integration in NVIDIA Dynamo-Triton
The Verge AI1 个来源

California tightens rules on AI data center energy and water use

California Gov. Gavin Newsom has signed seven bills designed to prevent AI data centers from passing utility costs onto residents, as reported earlier by the Los Angeles Times. The package of laws requires the California Public Utilities Commission to introduce a new rate classification for data centers while forcing them to pay for upgrades to […]

其他1 个来源
来源与时间线
The Verge AICalifornia tightens rules on AI data center energy and water use
llama.cpp1 个来源

b11081

<details open> test-llama-archs : make tensor data stdev configurable and improve help (#29133) * test-llama-archs : make tensor data stdev configurable Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * test-llama-archs : expand usage and add examples Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * test-llama-archs : fail on unknown args and log usage Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * test-llama-archs : add test run summary Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * test-llama-archs : initialize Mamba ssm_a negative Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-Vision-Exp * test-recurrent-state-rollback : report NMSE for logits mismatches As

其他1 个来源
来源与时间线
llama.cppb11081
TechCrunch AI1 个来源

Discover what’s next: 5 days left to save up to $200 on your TechCrunch Disrupt 2026 ticket

Five days left to save up to $200 on your TechCrunch Disrupt 2026 pass + 50% off a second one. Join 10,000+ founders, investors, and operators at San Francisco’s Moscone West, October 13-15. Grab your ticket savings before prices go up on September 25 at 11:59 p.m. PT.

其他1 个来源
来源与时间线
TechCrunch AIDiscover what’s next: 5 days left to save up to $200 on your TechCrunch Disrupt 2026 ticket

9月21日周一

The Verge AI1 个来源

Can John Ternus find Apple’s next big thing?

Today, I’m talking with Mark Gurman, who is the world’s best-sourced Apple reporter — both as Bloomberg’s chief Apple correspondent and the host of the upcoming podcast Power On. Earlier this month, Apple hosted its annual iPhone hardware event, virtually every detail of which was reported ahead of time by Mark. But this event was […]

其他1 个来源
来源与时间线
The Verge AICan John Ternus find Apple’s next big thing?
The Verge AI1 个来源

iPhone owners can now submit claims in Apple’s $250 million Siri AI settlement

Apple is paying $250 million to settle claims that it failed to deliver an AI-upgraded Siri - and now, eligible iPhone owners can submit a claim for a payout. If you live in the US and purchased an iPhone 15 Pro, iPhone 15 Pro Max, or any iPhone 16 model between June 10th, 2024, and […]

其他1 个来源
来源与时间线
The Verge AIiPhone owners can now submit claims in Apple’s $250 million Siri AI settlement
MIT Technology Review AI1 个来源

4 ways to address the failures we found along the US border’s “virtual wall”

MIT Technology Review today published our investigation into how many people have died near the “virtual wall” of surveillance towers that the US government has installed along the US-Mexico border. We found cases of people who walked undetected through areas surveilled by advanced, AI-enabled towers and later died nearby, where their bodies remained unnoticed for…

其他1 个来源
来源与时间线
MIT Technology Review AI4 ways to address the failures we found along the US border’s “virtual wall”
MIT Technology Review AI1 个来源

The US spent billions on border surveillance. Why can’t it catch people before they die?

When José Morales Bernal crossed the border into the United States on April 8, 2024, the day before his 32nd birthday, it should have triggered a chain of technological alerts and human responses. As he walked through the desert in southern New Mexico that morning, he was within range of three surveillance towers. Newly installed…

其他1 个来源
来源与时间线
MIT Technology Review AIThe US spent billions on border surveillance. Why can’t it catch people before they die?
llama.cpp1 个来源

b11065

<details open> CUDA: tune FA for Gemma 4 on Ampere or newer (#29152) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48802880> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [

其他1 个来源
来源与时间线
llama.cppb11065
The Verge AI1 个来源

No one is surprised that Nvidia’s Jensen Huang thinks AI fears are overblown.

The man who may stand to make the most money from the AI boom seems to think he knows better than anyone else, including researchers who have studied and worked on AI for decades. In an interview with CBS Sunday Morning, he claimed there was a "0% chance" of AI being the end of the […]

其他1 个来源
来源与时间线
The Verge AINo one is surprised that Nvidia’s Jensen Huang thinks AI fears are overblown.

9月20日周日

The Verge AI1 个来源

Trump now says he wants to form an ‘AI Force’

The president posted on Truth Social that he wanted to appoint an "AI czar" to lead a new "AI force." He made the announcement amid growing calls from across the political spectrum and even within the industry to pump the brakes on AI development. He posted that his administration "will not in any way hinder […]

其他1 个来源
来源与时间线
The Verge AITrump now says he wants to form an ‘AI Force’
llama.cpp1 个来源

b11064

<details open> metal : support arbitrary hc in dsv4_hc_pre (#29169) the dsv4_hc_pre kernels hardcoded hc = 4 via a constexpr used with simd_shuffle, so the op was rejected by supports_op for any other hc and fell back to CPU. Kimi-K3 uses dsv4_hc_pre with hc equal to the number of banked checkpoints in the cross-layer residual stack, which grows with the layer index. pass n_hc as a function constant (FC_DSV4_HC) with per-n_hc pipeline variants, and loop over it in both pre kernels with direct loads add test-backend-ops cases for hc = 1, 2, 3, 5, 8 and 65, gated and not gated Assisted-by: pi:llama.cpp/Qwen3.8-27B </details> **Website:** - <https://llama.app> **Attestations:** - <https://githu

其他1 个来源
来源与时间线
llama.cppb11064
llama.cpp1 个来源

b11063

<details open> common/peg : handle invalid utf-8 sequences in the AST (#29161) * common/peg : handle invalid utf-8 sequences in the AST * cont : return maximal subpart per Unicode recommendations * cont : remove strict argument </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48751230> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11063/lla

其他1 个来源
来源与时间线
llama.cppb11063
The Verge AI1 个来源

Humans, not rogue AI, are still the biggest cybersecurity risk to energy systems

Before recent high-profile hacks raised the specter of AI possibly "killing all humans," our energy systems were already disturbingly vulnerable to cyberattack - and the risk is growing. "We were always prey. We were just kind of surviving at the appetite of our predators," Joshua Corman, executive in residence for public safety and resilience at […]

其他1 个来源
来源与时间线
The Verge AIHumans, not rogue AI, are still the biggest cybersecurity risk to energy systems
llama.cpp1 个来源

b11057

<details open> chat : add dedicated Ling 3.0 (Bailing V3) parser (#28682) * chat: add dedicated Ling 3.0 (Bailing V3) parser Ling 3.0 Flash templates pre-open the think block in the generation prompt, so the model never emits an opening <think>, and a tool call can arrive before any </think>. The generated autoparser terminated reasoning only at the close tag, which classified such tool calls entirely as reasoning_content: clients received content="" with no tool_calls and agent loops died as reasoning-only turns. Adds a specialized parser that terminates reasoning at the think close tag or at a <tool_call> start, mirroring the hand-written Qwen3-Coder and Kimi K3 parsers and the reference v

其他1 个来源
来源与时间线
llama.cppb11057
The Verge AI1 个来源

Meta’s Muse is creepy, but maybe not for the reasons you think

Meta's Muse is apparently an effective AI assistant, but one that's a little creepy. Part of that is because of its new Mac app, which can access Messages, Calendar, and Notes. But for all its smarts, Muse doesn't actually know how to describe itself. Jason Aten, a contributing editor at Inc Magazine, posted on Threads […]

其他1 个来源
来源与时间线
The Verge AIMeta’s Muse is creepy, but maybe not for the reasons you think
llama.cpp1 个来源

b11056

<details open> hexagon: enable I32 GET_ROWS (#29116) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48662388> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)

其他1 个来源
来源与时间线
llama.cppb11056
llama.cpp1 个来源

b11055

<details open> hexagon: add support for GEGLU_QUICK (#29114) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48660160> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x

其他1 个来源
来源与时间线
llama.cppb11055

9月19日周六

The Verge AI1 个来源

Gemini went rogue, hacked three companies, and Google hid it

In May, Gemini broke containment and hacked three different companies, but Google didn't disclose the incident until the Wall Street Journal approached the company. The hacks happened during a test of the model's cybersecurity capabilities run by third-party Irregular, which was also involved in similar incidents involving Meta and OpenAI. According to WSJ, Google didn't […]

其他1 个来源
来源与时间线
The Verge AIGemini went rogue, hacked three companies, and Google hid it
The Verge AI1 个来源

Does AI need an antitrust exemption so it doesn’t kill everyone????

Today on Decoder, we’ve got the first of a two-part series on the future of business, and I’m talking with Jonathan Kanter, the former antitrust chief for the US Department of Justice in the Biden administration. These days, he’s both a professor of law at WashU and professor of technology policy at Carnegie Mellon. The […]

监管1 个来源
来源与时间线
The Verge AIDoes AI need an antitrust exemption so it doesn’t kill everyone????
Ollama1 个来源

v0.34.3

What's Changed `GET /api/show` now advertises each model's thinking controls and default: ```sh curl http://localhost:11434/api/show -d '{"model": "glm-5.3-flash:cloud"}' ``` ```json { "thinking": { "values": ["low", "high", "max"], "default": "max" } } ``` * **Nemotron H** vision models are now supported on Apple Silicon with MLX * Ollama's macOS app will now no longer reopen windows you've closed when activating it **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.2...v0.34.3-rc0

其他1 个来源
来源与时间线
Ollamav0.34.3
Ollama1 个来源

v0.34.3

What's Changed `GET /api/show` now advertises each model's thinking controls and default: ```sh curl http://localhost:11434/api/show -d '{"model": "glm-5.3-flash:cloud"}' ``` ```json { "thinking": { "values": ["low", "high", "max"], "default": "max" } } ``` * **Nemotron H** vision models are now supported on Apple Silicon with MLX * Ollama's macOS app will now no longer reopen windows you've closed when activating it * Fix for model pulls from HuggingFace **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.2...v0.34.3-rc0

其他1 个来源
来源与时间线
Ollamav0.34.3
Ollama1 个来源

v0.34.3

What's Changed `GET /api/show` now advertises each model's thinking controls and default: Available in the CLI with: ``` ollama show gemma4 ``` ``` thinking levels false, true default true ``` Available in the API with: ```sh curl http://localhost:11434/api/show -d '{"model": "glm-5.3-flash:cloud"}' ``` ```json { "thinking": { "values": ["low", "high", "max"], "default": "max" } } ``` Also available on ollama.com directly for cloud models. * **Nemotron H** vision models are now supported on Apple Silicon with MLX * Ollama's macOS app will now no longer reopen windows you've closed when activating it * Fix for model pulls from HuggingFace **Full Changelog**: https://github.com/ollama/ollam

其他1 个来源
来源与时间线
Ollamav0.34.3
llama.cpp1 个来源

b11046

<details open> opencl: add support for bin kernel `flash_attn_f32_f16_bin` (#29046) * opencl: add `flash_attn_f32_f16_bin` * opencl: guarded prefill fa </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48577094> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-

其他1 个来源
来源与时间线
llama.cppb11046
SGLang1 个来源

v0.5.20

Highlights *713 PRs from 237 contributors.* **New models in this release** (see the cookbook for all supported models): | Model | Type | PRs | Cookbook | |---|---|---|---| | GLM-5.3-Flash | Autoregressive | #36507, #38621 | link | | Hy4-Preview | Autoregressive | #36805 | link | | Qwen3.8-Flash-Next | Autoregressive | [#37500](https://github.com/sgl-project/sglang/pu

其他1 个来源
来源与时间线
SGLangv0.5.20
llama.cpp1 个来源

b11045

<details open> hexagon: add ROLL op support (#29105) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48565469> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)

其他1 个来源
来源与时间线
llama.cppb11045
llama.cpp1 个来源

b11044

<details open> hexagon: im2col update (#29103) * ggml-hexagon: accept 1D and padded IM2COL ops * ggml-hexagon: make pure-DDR IM2COL kernel is_2D-aware * ggml-hexagon: extend IM2COL DMA patch-embed fast path to 1D * ggml-hexagon: add blocked-staging general IM2COL DMA kernel </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48558320> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/

其他1 个来源
来源与时间线
llama.cppb11044
The Verge AI1 个来源

OpenAI and Microsoft knew they were starting a ‘doom loop’ for the web

Recently unsealed court documents in the New York Times' case against OpenAI and Microsoft are pretty damning. The companies' own documentation warned that it was starting a "doom loop" that would damage the web, characterized its scraping of data to train its models as the "largest theft of labor in human history," and that it […]

监管1 个来源
来源与时间线
The Verge AIOpenAI and Microsoft knew they were starting a ‘doom loop’ for the web
Microsoft Foundry Blog1 个来源

A2A Endpoints and A2A Tool in Microsoft Foundry agents | Microsoft Community Hub

Agent-to-agent collaboration just became much easier to build in Microsoft Foundry. The A2A Tool and incoming A2A endpoints support A2A protocol version 1.0,...

其他1 个来源
来源与时间线
Microsoft Foundry BlogA2A Endpoints and A2A Tool in Microsoft Foundry agents | Microsoft Community HubMicrosoft Foundry BlogBuild a SAFE Agent on Microsoft Foundry | Microsoft Community HubMicrosoft Foundry BlogBuilding Production-Ready AI Agents in Microsoft Foundry: 10 Lessons Learned | Microsoft Community Hub
The Verge AI1 个来源

Virginia governor creates an AI task force and moves to restrain data centers

Virginia Gov. Abigail Spanberger ordered the state government to take steps that could empower local communities to have a larger say in data center development and slow down approvals in a state that is already home to the data center capital of the world. Executive Order 22 bans executive branch officials from signing nondisclosure agreements […]

其他1 个来源
来源与时间线
The Verge AIVirginia governor creates an AI task force and moves to restrain data centers
The Verge AI1 个来源

Gavin Newsom is pushing for an AI kill switch

California Gov. Gavin Newsom (D) is positioning the state to take the lead on AI oversight, including the potential to mandate a "kill switch" for frontier models, with a new executive order issued Friday. Newsom's order directs the state to convene a group of experts that will deliver recommendations within two months on how to […]

监管1 个来源
来源与时间线
The Verge AIGavin Newsom is pushing for an AI kill switch
Anthropic Python SDK1 个来源

v1.7.0

1.7.0 (2026-09-18) Full Changelog: v1.6.0...v1.7.0 Features * **api:** add group with display_name to rate limits, deprecate group_type (28f0a83) * **tools:** add compact_before_next_turn() to the tool runner (#641) (8b23fb3) Bug Fixes * **bedrock:** raise an API error for eventstream exception and error frames ([#769](https://github.com/an

监管1 个来源
来源与时间线
Anthropic Python SDKv1.7.0

9月18日周五

OpenAI Python SDK1 个来源

v3.16.0

3.16.0 (2026-09-18) Features * **api:** add webhook endpoint management (#3892) (9a11f6e) Chores * **api:** deprecate MCP connector_id (#3894) (1ccaf07)

监管1 个来源
来源与时间线
OpenAI Python SDKv3.16.0
OpenAI Python SDK1 个来源

v3.15.0

3.15.0 (2026-09-18) Features * **api:** add agent session model settings (#3882) (4b15817) * **api:** add audio-mini model choices (#3886) (a6eeb3f) * **api:** add compaction progress events (#3866) ([98e1d24](https://github.com/openai/openai-python/commit/98e1d24f4902ab58830a

监管1 个来源
来源与时间线
OpenAI Python SDKv3.15.0
TechCrunch AI1 个来源

Google DeepMind launches institute to widen the AGI debate

The new institute aims to surface differing views between Google, Google DeepMind, and the broader global research community around AGI. "They will not always agree, and they will likely change their minds, as more data and information comes to light at the fast-moving frontier."

其他1 个来源
来源与时间线
TechCrunch AIGoogle DeepMind launches institute to widen the AGI debate
The Verge AI1 个来源

Claude Code relaunches Projects to manage multiple AI agents in the cloud

The revamped projects feature in Claude Code allows users to run multiple agents under the same roof, with a shared memory, goals, and library of files and artifacts. Similar to Grok Bot and other tools that manage groups of AI agents, each project has "threads" running different tasks in parallel, with a "coordinator" directing everything: […]

其他1 个来源
来源与时间线
The Verge AIClaude Code relaunches Projects to manage multiple AI agents in the cloud
Microsoft Foundry Blog1 个来源

Announcing the Playwright Workspaces Remote MCP Server for Agentic Browser Automation | Microsoft Community Hub

An agent can analyze a request, choose the right next step, and call every available API—then stall when the final action exists only in a web interface. AI...

其他1 个来源
来源与时间线
Microsoft Foundry BlogAnnouncing the Playwright Workspaces Remote MCP Server for Agentic Browser Automation | Microsoft Community Hub
llama.cpp1 个来源

b11028

<details open> ci : add missing evict-old-files (#29041) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48252257> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (

其他1 个来源
来源与时间线
llama.cppb11028
llama.cpp1 个来源

b11027

<details open> rpc : skip ACCEL devices (#29020) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48232585> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](ht

其他1 个来源
来源与时间线
llama.cppb11027

9月17日周四

The Verge AI1 个来源

Microsoft AI CEO says AI threats are real, and Anthropic is making it worse

Today, I’m talking with Mustafa Suleyman, the CEO of Microsoft AI. As you’re no doubt aware, the biggest story in tech right now is the spiraling debate about AI safety and regulation. It should come as no surprise that Mustafa has strong opinions on how AI should be built and regulated. Microsoft just published a […]

其他1 个来源
来源与时间线
The Verge AIMicrosoft AI CEO says AI threats are real, and Anthropic is making it worse
llama.cpp1 个来源

b11026

<details open> model : skip gate_up_exps if TENSOR_SKIP is set (#29014) required for qwen35moe if MTP tensors are fused but not loaded </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48194938> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/rel

其他1 个来源
来源与时间线
llama.cppb11026
llama.cpp1 个来源

b11010

<details open> vulkan: work around NV bug with argsort_large.comp (#28975) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48059150> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:

其他1 个来源
来源与时间线
llama.cppb11010
The Verge AI1 个来源

Snap is launching a new Specs AI tool, and it’s coming to iOS and Mac

Snap is introducing "Specs Intelligence," a new AI assistant that can connect other digital accounts to help you with things like work tasks and keeping track of travel information. It seems similar to AI assistants like Meta's Muse and Gemini's Spark, though Snap is pitching Specs Intelligence as an "anticipatory AI service" that "helps you […]

其他1 个来源
来源与时间线
The Verge AISnap is launching a new Specs AI tool, and it’s coming to iOS and Mac
TechCrunch AI1 个来源

Anthropic and OpenAI want to embed safety evaluators. Will they really be independent?

Anthropic and OpenAI want to embed independent safety evaluators inside their AI labs. Researchers welcome the unprecedented access, but warn meaningful oversight requires transparency, independence, and eventually regulation.

其他1 个来源
来源与时间线
TechCrunch AIAnthropic and OpenAI want to embed safety evaluators. Will they really be independent?
The Verge AI1 个来源

The 2.5-hour AI-generated Odyssey movie is 2.5 hours too long

Christopher Nolan's engrossing take on The Odyssey dominated at the box office and spurred a newfound interest in classic literature among filmgoers. But a new retelling of the story made entirely with AI is so bad that it might just make viewers hate the original tale altogether. The new film, called Odysseus: The Fall, comes […]

其他1 个来源
来源与时间线
The Verge AIThe 2.5-hour AI-generated Odyssey movie is 2.5 hours too long
The Verge AI1 个来源

The AI data center e-waste problem is huge — and getting bigger

E-waste from the AI boom has been vastly underestimated, a new report warns. By 2050, it could become enough trash to fill 23 million shipping containers - roughly enough 40-foot containers to circle the world six times if lined up in a row. It's a significantly higher estimate of AI's e-waste than previous studies have […]

其他1 个来源
来源与时间线
The Verge AIThe AI data center e-waste problem is huge — and getting bigger
NVIDIA Generative AI Technical Blog1 个来源

TensorRT Edge-LLM Completes the MLPerf Edge Agentic Benchmark 6.4x Faster on Jetson AGX Thor

AI agents are moving from cloud data centers to vehicles, robots, and other edge devices. Unlike a chatbot that answers a single prompt, an agent works through...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogTensorRT Edge-LLM Completes the MLPerf Edge Agentic Benchmark 6.4x Faster on Jetson AGX Thor
llama.cpp1 个来源

b11009

<details open> TP: fix split state and granularity for fused QKV gemma4, qwen35 (#28965) * model: calculate split states for attn_qkv from n_head * n_embd_head_k required for gemma4 with --fuse-qkv, where n_embd is 5376 but Q is 8192. * model: handle fused full attention layers for qwen35/qwen35moe * model: add TODO: [TAG_SPLIT_QGATE_QWEN] </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/48008214> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-

其他1 个来源
来源与时间线
llama.cppb11009
llama.cpp1 个来源

b11007

<details open> Enable CUDA graph for MTP draft (#28549) * Improve CUDA graph usage for MTP * Rename field * Address review feedback </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47982528> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releas

其他1 个来源
来源与时间线
llama.cppb11007

9月16日周三

MIT Technology Review AI1 个来源

Building the materials foundation for AI

The AI boom is becoming a materials challenge. As AI pushes computing into new territory, the materials behind that infrastructure are becoming just as crucial as the algorithms running on it. Semiconductors and data centers are approaching physical limits around performance, thermal management, electrical efficiency, and reliability, creating new demands for materials that can do…

其他1 个来源
来源与时间线
MIT Technology Review AIBuilding the materials foundation for AI
TechCrunch AI1 个来源

We don’t need AI regulation — leave safety to us, Nvidia’s Jensen Huang says

AI isn't some kind of new form of "alien mind," according to Jensen Huang. It's just hardware and software, so safety can be engineered by each AI product maker.

其他1 个来源
来源与时间线
TechCrunch AIWe don’t need AI regulation — leave safety to us, Nvidia’s Jensen Huang saysTechCrunch AINvidia CEO Jensen Huang tells Trump ‘we’re not going to let [an AI slowdown] happen’
llama.cpp1 个来源

b10991

<details open> hexagon: add back missing contiguous fast-path and hvx_copy_uu for each run (#28886) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47775478> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10991/llama-b10991-

其他1 个来源
来源与时间线
llama.cppb10991
The Verge AI1 个来源

AI and data centers are incredibly unpopular in every poll

Poll data released Tuesday by the New York Times and Siena University confirms what we've already been seeing, and what politicians are responding to - AI and data centers are incredibly unpopular. Asked if they support or oppose the construction of data centers to power AI tech, 61 percent of the 1,503 likely voters surveyed […]

其他1 个来源
来源与时间线
The Verge AIAI and data centers are incredibly unpopular in every poll
llama.cpp1 个来源

b10990

<details open> hex-cpy: use dma if src and dst are contiguous (#28906) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47770889> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** -

其他1 个来源
来源与时间线
llama.cppb10990
OpenAI Python SDK1 个来源

v3.14.1

3.14.1 (2026-09-15) Bug Fixes * **client:** validate retry limits and preserve application errors (#3867) (f86c721) * correct typo "th" to "the" in StreamAlreadyConsumed error message (#3022) (7186203) * **examples:** correct Azure endpoint hostname (#3298) ([543516c](https://

监管1 个来源
来源与时间线
OpenAI Python SDKv3.14.1
llama.cpp1 个来源

b10989

<details open> HIP: Enable AllReduce for ROCm (#27825) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47762387> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CP

其他1 个来源
来源与时间线
llama.cppb10989
Ollama1 个来源

v0.34.2

What's Changed * llama.cpp updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.1...v0.34.2-rc0

其他1 个来源
来源与时间线
Ollamav0.34.2
Ollama1 个来源

v0.34.2

What's Changed - Added first-run setup when running `ollama`, with options to sign in or continue locally. Setup completion is shared with the desktop app on macOS and Windows. - Added `ollama://apps` to open the desktop app’s Apps page directly on macOS and Windows. - Fixed excessive memory growth during long generations with MLX speculative decoding. - Updated llama.cpp. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.1...v0.34.2

其他1 个来源
来源与时间线
Ollamav0.34.2
Ollama1 个来源

v0.34.2

What's Changed * llama.cpp updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.1...v0.34.2-rc0

其他1 个来源
来源与时间线
Ollamav0.34.2
NVIDIA Generative AI Technical Blog1 个来源

How NVIDIA Groq 3 LPX Deterministic Execution Drives Power-Efficient High-Interactivity Inference on NVIDIA Vera Rubin

Power is a defining constraint for AI factories. As AI workloads demand a full compute platform to serve them, each component of that platform must maximize...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogHow NVIDIA Groq 3 LPX Deterministic Execution Drives Power-Efficient High-Interactivity Inference on NVIDIA Vera RubinNVIDIA Generative AI Technical BlogHow NVIDIA Groq 3 LPX Unlocks Ultrafast Interactivity at Long Context on NVIDIA Vera Rubin

9月15日周二

Hugging Face PEFT1 个来源

v0.21.0

Highlights <img width="1248" height="560" alt="peft-v0 21 0" src="https://github.com/user-attachments/assets/c2b8b162-6564-4c7f-b5b6-130b8808ea18" /> New methods Riemannian-preconditional LoRA optimizer Riemannian-preconditioned LoRA optimizer (#3382), contributed by @smellslikeml, is a new LoRA-specific optimizer similar to LoRA-FA or LoRA+ that applies a Riemannian pre-conditioning on the gradients to improve feature learning and avoiding separate tuning of learning rates for LoRA's A and B matrices. LoRA va

其他1 个来源
来源与时间线
Hugging Face PEFTv0.21.0
Anthropic Python SDK1 个来源

v1.6.0

1.6.0 (2026-09-15) Full Changelog: v1.5.0...v1.6.0 Features * **api:** add auto mode tool permissions for Managed Agents (909d92f) * **api:** add compaction parameter and signed compaction blocks (beta) (8689179) * **api:** add enum types for workspace data-residency geo fields (3dc6dbf) * **api:** add think

其他1 个来源
来源与时间线
Anthropic Python SDKv1.6.0
The Verge AI1 个来源

Meta’s new One subscriptions put a price on social media and AI

Shortly after launching its new do-everything AI assistant Muse, Meta's launching subscription bundles that pair its standalone app subscriptions with extra AI usage. Some of the new Meta One bundles were in testing earlier this year, but are now available globally starting today, with several tiers for individual users, creators, and businesses. Meta says the […]

其他1 个来源
来源与时间线
The Verge AIMeta’s new One subscriptions put a price on social media and AI
The Verge AI1 个来源

This doorbell camera lets a human security guard watch your front door

DIY home security company SimpliSafe is bringing its AI-powered proactive security feature to the front door. The new SimpliSafe Video Doorbell Series 2 launches today for $199.99 and works with the company's Active Guard Outdoor Protection (starting at $49.99 a month). This combines AI analysis with live agents to detect potential threats and respond proactively. […]

其他1 个来源
来源与时间线
The Verge AIThis doorbell camera lets a human security guard watch your front door
OpenAI Python SDK1 个来源

v3.14.0

3.14.0 (2026-09-14) Features * **streaming:** normalize errors raised while reading streams (#3827) (d7c41ef) Bug Fixes * bound vector store file polling (#3401) (ae41bc4) * **example:** refresh realtime push_to_talk_app session types (#2926) ([0b7ad38](https://github.com/

监管1 个来源
来源与时间线
OpenAI Python SDKv3.14.0
The Verge AI1 个来源

Is Big Tech’s AI slowdown a safety pact or a cartel?

When OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, Google DeepMind cofounder Demis Hassabis, and SpaceX head Elon Musk loosely agreed over the weekend to slow down AI development, skeptics spotted an ulterior motive immediately. The AI titans had declared that their aim was to "pace the frontier," signing on at least partially to a […]

其他1 个来源
来源与时间线
The Verge AIIs Big Tech’s AI slowdown a safety pact or a cartel?
Ollama1 个来源

v0.34.1

What's Changed * app: fix ChatGPT model selector spacing * mlxrunner: Evict prefix cache snapshots from the active conversation * mlxrunner: check system free memory and wait for evicted runners before loading the next MLX model * llm: raise token repeat limit to 100 and return error instead of incomplete result * mlx: scope array lifetimes instead of pinning and sweeping * llm: keep gemma3n projector off the CPU * app: refresh Apps layout and command copy feedback * MLX and llama.cpp updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.0...v0.34.1-rc1

其他1 个来源
来源与时间线
Ollamav0.34.1
Ollama1 个来源

v0.34.1

What's Changed * MLX safetensors `ollama create` no longer experimental. GGUF model creation now requires using llama.cpp tooling for safetensor conversion and quantization. * Improved MLX memory handling on Apple Silicon * Runaway repeat token detection now requires 100 repeat tokens for reduced false positives (e.g. OCR) * `/api/tags` is much faster on large model libraries (3.1 s → 294 ms cold in testing), and model capabilities are now reported consistently. * Deprecated `typical_p`: it can no longer be set when creating new models, existing GGUF models retain support. * MLX and llama.cpp updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.34.0...v0.34.1-rc1

其他1 个来源
来源与时间线
Ollamav0.34.1
The Verge AI1 个来源

What execs and politicians are saying about slowing down AI development

Dario Amodei kicked off a flood of statements over the past few days about AI safety by publishing a long essay titled "We Must Pace the Frontier" detailing why AI development should be slowed down. Other AI leaders and politicians are speaking out in favor of or opposing his points, and we've compiled some of […]

其他1 个来源
来源与时间线
The Verge AIWhat execs and politicians are saying about slowing down AI development
llama.cpp1 个来源

b10970

<details open> HIP: fattn-mma: use fp32 accumulation on MFMA devices (#28576) use fp32 accumulators in fattn-mma on CDNA </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47450174> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download

其他1 个来源
来源与时间线
llama.cppb10970
The Verge AI1 个来源

Jensen Huang puts Trump on speakerphone onstage to announce robots won’t take over the world

Nvidia CEO Jensen Huang took a call from President Trump on Monday while onstage at the All-In Podcast's All-In Summit. It's not the first time Huang has taken a call from the president during work, but this time he put Trump on speakerphone before a big crowd. During the call, the president launched into his […]

其他1 个来源
来源与时间线
The Verge AIJensen Huang puts Trump on speakerphone onstage to announce robots won’t take over the world
llama.cpp1 个来源

b10969

<details open> ci : add ubuntu-cuda builds to release (#28186) * release : add ubuntu-cuda build job (12.8/13.3, x64+arm64) * Add GCC 14 for CUDA arm64 builds in CI * Eplicit bash * Install git for CCCL fetch * Install git before we clone/checkout * Match CI names for WIndows * Whitelist llama.cpp repo to git * Use $GITHUB_WORKSPACE * Also ship dependent libs on Ubuntu Need NCCL additionally as it's pre-built available on Linux * Avoid duplicate files in packaged cudart * Copy NCCL license * Install CURL to fetch NCCL license * Update .github/workflows/release.yml Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> * Remove NCCL until licensing has been confirmed --------- Co

其他1 个来源
来源与时间线
llama.cppb10969
llama.cpp1 个来源

v0.4.1

Overview llama.cpp 0.4.1 adds Maple 20B-A1B, Tencent Hy 4, and Spark2.5 support, improves JSON schema handling, chat parsing, logging, and server child-process management, and updates ggml to v0.24.0. API changes - Changed `llama_sampler_chain_n()` to return `int32_t` instead of `int` (#28631). - Added `server_subproc` and `waiter` for router child-process monitoring in `server-common.h` (#28555). New models - Added Maple 20B-A1B ternary MoE architecture (CPU) (#27000). - Added Tencent Hy 4 (`hy_v4`) preview architecture suppor

其他1 个来源
来源与时间线
llama.cppv0.4.1
MIT Technology Review AI1 个来源

The AI industry has taken a doomer turn. What now?

This story appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. This weekend, Dario Amodei, CEO of Anthropic, posted an essay calling for a brake on the pace of development of LLMs. Amodei cites the looming dangers he sees from the technology, from its…

其他1 个来源
来源与时间线
MIT Technology Review AIThe AI industry has taken a doomer turn. What now?
Google Gemini Blog1 个来源

DevFest is back

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/DevFest.max-600x600.format-webp.webp">DevFest 2026 is back and here’s how you can connect with one of the more than 800 global events to build, secure, and scale in the agentic AI era.

其他1 个来源
来源与时间线
Google Gemini BlogDevFest is back
MIT Technology Review AI1 个来源

AI agents blew the whistle on their cheating colleagues

A group of AI agents asked to solve a series of math problems split into rival factions—when some cheated, others tried to stop them. That whistleblowing behavior, seen for the first time in a recent experiment run by Google DeepMind, could have implications for alignment researchers trying to keep swarms of autonomous AI agents in…

其他1 个来源
来源与时间线
MIT Technology Review AIAI agents blew the whistle on their cheating colleagues

9月14日周一

Microsoft Foundry Blog1 个来源

Can a Harmless Prompt Break an AI Guardrail? | Microsoft Community Hub

Tamara Gaidar, Senior Data Scientist, Microsoft Security Research Team Maor Nissan, Principal Data Scientist, Microsoft Security Research Team Where we left...

其他1 个来源
来源与时间线
Microsoft Foundry BlogCan a Harmless Prompt Break an AI Guardrail? | Microsoft Community HubMicrosoft Foundry BlogCan a Harmless Prompt Break an AI Guardrail? | Microsoft Community Hub
The Verge AI1 个来源

Trump and Mike Johnson think the AI industry is overreacting

Yesterday, Anthropic CEO Dario Amodei published a lengthy open letter saying it was time to "pace the frontier" and slow down AI development. OpenAI's Sam Altman and Elon Musk both agreed, publicly voicing their support on X. Even Alphabet's Demis Hassabis offered tentative support for Amodei's proposal. Donald Trump and House Speaker Mike Johnson, however, […]

其他1 个来源
来源与时间线
The Verge AITrump and Mike Johnson think the AI industry is overreacting
llama.cpp1 个来源

b10948

<details open> tests : exclude HY_V4 from WebGPU test-llama-archs tests (#28855) Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47194535> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/r

其他1 个来源
来源与时间线
llama.cppb10948
llama.cpp1 个来源

b10947

<details open> models : guard the expert FFN size fallback in nemotron-h against a zero divisor (#28779) The NextN/MTP tail loop derives the expert FFN size as n_ff/n_expert_used when expert_feed_forward_length gives nothing for the layer. Both values come from per-layer arrays that legitimately hold 0 on layers that are not MoE, so a checkpoint whose predict layers hold 0 in both divides by zero and dies with SIGFPE at load time, with no error message. Report the malformed metadata instead. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47192495> **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/

其他1 个来源
来源与时间线
llama.cppb10947
llama.cpp1 个来源

b10946

<details open> ggml-cpu(s390x): guard VXE-only repack helpers (#28775) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47190488> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** -

其他1 个来源
来源与时间线
llama.cppb10946

9月13日周日

llama.cpp1 个来源

b10935

<details open> common: add LOG_JSON macro to log structured data (#28586) * add LOG_JSON macro * fit: add demo LOG_JSON </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47112879> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/

其他1 个来源
来源与时间线
llama.cppb10935
llama.cpp1 个来源

b10934

<details open> common : implement common_schema internal representation for JSON schemas (#28736) * common : implement common_schema types * common : implement a json schema optimizer * common : reduce optimizations * common : refactor json-schema-to-grammar to use common_schema * common : use common_trie * common/schema : implement type/kind resolution * cont : cleanup * cont : remove common_chat_tool_parameters * cont : simplify schema resolution * cont : pass common_schema through the json-schema-to-grammar builder * cont : cleanup * cont : move enums under common_schema and add type enum * cont : reduce test cases * cont : clean up * cont : clean up * refactor : rename common_schema_pars

其他1 个来源
来源与时间线
llama.cppb10934
llama.cpp1 个来源

b10933

<details open> jinja : support dot property integer literals (#28817) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/47107718> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** -

其他1 个来源
来源与时间线
llama.cppb10933
The Verge AI1 个来源

OpenAI’s rogue AI tried to hack another company in May

In May, hundreds of malicious and spam packages were uploaded to RubyGems, causing a serious disruption for the host. Now independent researchers have said that a swarm of OpenAI agents were responsible for the attack. Not only that, but the AI tried to steal users' API keys. At the time, RubyGems described it as a […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s rogue AI tried to hack another company in May
The Verge AI · TechCrunch AI2 个来源

Sam Altman says OpenAI going public in 2026 would be ‘ill-advised’

OpenAI CEO Sam Altman confirmed that there would be no OpenAI IPO in 2026 during an interview with Fortune. Over the course of 45 minutes, Altman discussed a variety of subjects including the Hugging Face hacking incident, recursive self-improvement, and the possibility of building an AI that was beyond human control. On the latter, he […]

其他2 个来源
来源与时间线
The Verge AISam Altman says OpenAI going public in 2026 would be ‘ill-advised’TechCrunch AIOpenAI’s Sam Altman says it would be ‘ill-advised’ to go public in 2026
The Verge AI1 个来源

Anthropic CEO says it’s time to pump the brakes on AI

Anthropic CEO Dario Amodei says the time has come to slow down AI development and will give third-party evaluators like METR access to its models to help ensure its "adherence to safety practices and commitments." In a winding essay, Amodei proposed a three-step plan to "pace the frontier" - jargon that simply means to slow […]

其他1 个来源
来源与时间线
The Verge AIAnthropic CEO says it’s time to pump the brakes on AI

9月12日周六

TechCrunch AI1 个来源

Y Combinator’s Garry Tan wants US open-weight AI labs to ‘distill’ frontier models, too

Tan wants smaller, American open-weight AI labs to use the same kind of training techniques on American frontier AI labs, giving the U.S. a more robust set of open-weight options that aren’t Chinese.

其他1 个来源
来源与时间线
TechCrunch AIY Combinator’s Garry Tan wants US open-weight AI labs to ‘distill’ frontier models, too
The Verge AI1 个来源

Lawyer fined $5K over AI-hallucinated witnesses in a murder case

New Mexico's Supreme Court is punishing a lawyer for including AI-fabricated witnesses and fake police testimony in an appeal for his client's murder conviction, according to a report from Reuters. In a filing on Wednesday, the court fined Stephen Aarons $5,000 and held him in contempt for failing to "verify the factual claims and legal […]

监管1 个来源
来源与时间线
The Verge AILawyer fined $5K over AI-hallucinated witnesses in a murder case
llama.cpp1 个来源

b10917

<details open> cmake : skip PCH for llama-server PCH when using MSVC (#28763) This commit fixes an issue that I introduced when adding PCH (precompiled headers) in Commit 3bcfeb700fce9ff38a050dcd3f6a856319e948ba ("cmake : add PCH and unity build to improve build times (#28091)". See linked issue for details. Co-authored-by: mjungnickel18 Co-authored-by: Pascal <admin@serveurperso.com> Resolves: https://github.com/ggml-org/llama.cpp/issues/28758 Refs: https://github.com/ggml-org/llama.cpp/actions/runs/34592933983/job/103262608990#step:9:1284 </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46958058> **macOS/iOS:** - [macOS A

监管1 个来源
来源与时间线
llama.cppb10917
MIT Technology Review AI1 个来源

Roundtables: AI’s apocalypse crisis

Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? Join MIT Technology Review executive editor Niall Firth for a conversation with senior AI editor Will Douglas Heaven and AI reporter Grace Huckins unpacking AI extinction…

其他1 个来源
来源与时间线
MIT Technology Review AIRoundtables: AI’s apocalypse crisis
The Verge AI1 个来源

Anthropic spent this week in hot water over cybersecurity

After admitting earlier this year that its AI models had hacked other companies' systems on a handful of occasions, Anthropic released a new report on Wednesday detailing the attacks. It reveals a string of incidents displaying what Anthropic deems its models' single-minded "recklessness" - and will likely fuel already raging concerns about cybersecurity and AI. […]

其他1 个来源
来源与时间线
The Verge AIAnthropic spent this week in hot water over cybersecurity

9月11日周五

The Verge AI1 个来源

Meta says it’s changing AI suggestions after posing invasive personal questions

Meta says it's making changes to the prompts suggested by its AI chatbot after a viral video showed it digging for personal information about a woman's young daughters, as reported earlier by Futurism. In a statement to The Verge, Meta spokesperson Dina El-Kassaby says the company "missed the mark," adding that "the feature never should […]

其他1 个来源
来源与时间线
The Verge AIMeta says it’s changing AI suggestions after posing invasive personal questions
OGX1 个来源

v1.4.0

What's Changed * chore: bump fallback_version to 1.3.1.dev0 by @cdoern in https://github.com/ogx-ai/ogx/pull/6384 * fix(starter): default trust_remote_code to false for sentence-transformers by @yashanil98 in https://github.com/ogx-ai/ogx/pull/6318 * feat: Add Meta AI to ogx go auto-detection by @mattf in https://github.com/ogx-ai/ogx/pull/6383 * feat: add local_api_key auth provider with attributes and startup validation by @mattf in https://github.com/ogx-ai/ogx/pull/6400 * feat: default ogx go to localhost instead of all interfaces by @mattf in https://github.com/ogx-ai/ogx/pull/6402 * fix(file-processor): Preserve content in async Docling uploads by @pgustafs in https://github.com/ogx

其他1 个来源
来源与时间线
OGXv1.4.0
llama.cpp1 个来源

b10909

<details open> metal : single-source fusion table + fusion debug rework (#28164) * metal : rework fusion patterns into a single table All fusable op patterns for the Metal backend are now declared once in a fusion table (ggml-metal-fuse.cpp) and consumed by both the graph optimizer (ggml_metal_fuse_max, packing) and the op encoders (ggml_metal_fuse_next, compute). The two phases share the same pattern table plus ggml_can_fuse_subgraph_ext for the structural checks, and differ only in the mode used for the pattern check (STRUCTURAL at optimize time, since tensors are not allocated yet, and FULL at compute time, including Metal buffer placement). This also protects the snake refresh (MUL +

其他1 个来源
来源与时间线
llama.cppb10909
llama.cpp1 个来源

b10908

<details open> metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 (#28692) * metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 Generalize the row split from #28086 to the six other kernels that use the same lane-to-block mapping: iq1_s, iq1_m, iq2_xxs, iq2_xs, iq2_s and iq3_s. Each of them assigns one 32-element chunk per thread, so when a row has fewer than 32 chunks the rest of the simdgroup is idle. When nb32 < 32 and nb32 divides 32, 32/nb32 threads now share each chunk and each takes a slice of the rows, reusing the FC_mul_mv_split function constant and the dispatch wrapper introduced for iq3_xxs. The plain path is untouched: wide matri

其他1 个来源
来源与时间线
llama.cppb10908
llama.cpp1 个来源

b10901

<details open> vulkan: use CPU writes in ggml_backend_vk_cpy_tensor_async if the context is idle (#28618) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46709648> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10901/llama-b

其他1 个来源
来源与时间线
llama.cppb10901
The Verge AI1 个来源

Slack can now vibe-code interactive charts and reports inside chats

A new feature coming to Slack will allow you to build interactive reports, polls, dashboards, presentations, microsites, and other tools directly inside a chat. With Slackforce Surfaces, you can describe to Slackbot what you need, and it will use AI to gather information from relevant conversations and connected apps, like Google Drive or Salesforce, to […]

其他1 个来源
来源与时间线
The Verge AISlack can now vibe-code interactive charts and reports inside chats
The Verge AI1 个来源

Schools are catching on to Big Tech’s playbook

It's the hot new thing in tech, and it's where all the jobs are. Students who don't learn to use it fall behind. And to help them catch up in time, its creators are graciously providing the resources and curriculum for learning it, often pro bono. That's the narrative AI companies are pitching schools on […]

其他1 个来源
来源与时间线
The Verge AISchools are catching on to Big Tech’s playbook
OGX1 个来源

v1.0.3

This maintenance release for the 1.0 line addresses JOSS review findings and includes previously backported security, storage, and provider fixes. - Enforce configured route restrictions when creating batches, preserving the submitting caller's identity and provider credentials. #6522 - Update UI production dependencies to resolve the reported audit findings. #6519 - Enable UI TypeScript checks and repair the 1.0 playground's Responses, saved-chat, and file-search behavior. #6520 - Repair internal and GitHub documentation links and enforce strict documentation buil

其他1 个来源
来源与时间线
OGXv1.0.3
llama.cpp1 个来源

b10900

<details open> vulkan: use add_alloc_dep to enable topk_moe fusion for prefill (#28422) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46661423> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10900/llama-b10900-xcframework.

其他1 个来源
来源与时间线
llama.cppb10900
llama.cpp1 个来源

b10899

<details open> vulkan: small M matrix optimizations for qwen (#28457) * vulkan: optimize m=1 mul_mat by swapping A/B * vulkan: Improve small M perf Allow split_k with small M. Make small vs med tile selection (for coopmat2) depend on M, not just N. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46654773> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/release

其他1 个来源
来源与时间线
llama.cppb10899
Anthropic Python SDK1 个来源

v1.5.0

1.5.0 (2026-09-10) Full Changelog: v1.4.0...v1.5.0 Features * **api:** add auto mode tool permissions for Managed Agents (62aa21b) * **api:** add content_too_large web_fetch tool error code (4b5dec6) * **api:** add the user-profiles-2026-09-04 beta value and external_user_details to user profiles (5f881a5) *

其他1 个来源
来源与时间线
Anthropic Python SDKv1.5.0
OpenAI Python SDK1 个来源

v3.12.0

3.12.0 (2026-09-10) Features * **api:** Add Live API (0e4bfef) Bug Fixes * add aclose() to AsyncStream for standard async cleanup (#2854) (802b334) * handle bare `dict` and `list` annotations without type arguments (#3760) (c7e8c03) * pr

监管1 个来源
来源与时间线
OpenAI Python SDKv3.12.0

9月10日周四

The Verge AI1 个来源

Universal Music is launching an AI music platform with ElevenLabs

Universal Music Group is launching a new AI-powered platform that will allow users to draw from its catalog of licensed music to create song remixes, mashups, and new takes on tracks, according to an announcement on Thursday. The record label is developing the platform through a multiyear licensing agreement with ElevenLabs, a company that specializes […]

其他1 个来源
来源与时间线
The Verge AIUniversal Music is launching an AI music platform with ElevenLabs
MIT Technology Review AI1 个来源

Powering AI is an architecture problem

On July 22, 2026, a transmission line fault in Ashburn, Virginia—the heart of the world’s largest data center cluster—knocked more than 3 gigawatts of load off the grid in seconds. And it wasn’t the first time. Two years earlier, a single failed surge arrester dropped roughly 60 Virginia facilities and 1,500 megawatts at once. No…

其他1 个来源
来源与时间线
MIT Technology Review AIPowering AI is an architecture problem
NVIDIA Generative AI Technical Blog1 个来源

From Wafer-Out to First [redacted credential] Supply Chain Expertise with Nemotron and Palantir Foundry

NVIDIA has one of the largest and most complex supply chains in the world, and its performance is measured from wafer-out to first token. The interval is in two...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogFrom Wafer-Out to First [redacted credential] Supply Chain Expertise with Nemotron and Palantir Foundry
The Verge AI1 个来源

Suno releases its first AI music model made with record industry help

Suno's new v6 AI music model is its first made with support from the record industry. Suno's Jack Brody told The Verge that v6 was "trained from the ground up, with a new set of data that does not include the same data that our previous models were trained on." The data includes content licensed […]

其他1 个来源
来源与时间线
The Verge AISuno releases its first AI music model made with record industry help
The Verge AI1 个来源

OpenAI’s sly mathematical breakthrough sends a chill through academia

OpenAI's announcement Tuesday that it has solved one of mathematics' legendary Millennium Prize problems should have been a moment of triumph. The result is both an undeniable achievement and a striking demonstration of just how rapidly AI is transforming mathematics. But before it was even formally announced, the breakthrough had been complicated by the unusual […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s sly mathematical breakthrough sends a chill through academia
The Verge AI1 个来源

Read the Apple document explaining how new listening features still protect your privacy

At Wednesday's iPhone Duo launch event, Apple announced a handful of new Siri AI Audio Intelligence features, including Siri Recap, Live Rewind, Sound Recognition, and Music Recognition. Alongside its announcement, Apple released a document laying out how it plans to balance AI "ambient listening" and users' privacy. It says the raw audio from the new […]

其他1 个来源
来源与时间线
The Verge AIRead the Apple document explaining how new listening features still protect your privacy
llama.cpp1 个来源

b10883

<details open> vulkan: use spec constant for matrix matrix multiplication A-type (#25773) * vulkan: use spec constant for mul mat type_a vulkan: use map for mul_mm shapes cleanup fix indentation fix cm2 and shmem init fix cm2 spec constants fix cm2 bindings consolidate shmem tables and reduce size by type spec constant fix compiler warning fix missing Q2_0 type fix unused warning when integer dot glslc support is missing use minimal shmem size 8 instead of 1 to workaround cm2 compiler bug fix missing Q2_0 type in cm2 matmul fix types * remove LUT quants from unified shader * clean up * restore coopmat2 q4_k/q5_k optimization * split out q4_k/q5_k cm2 shader to fix Ampere regression * revert

其他1 个来源
来源与时间线
llama.cppb10883

9月9日周三

Hugging Face Transformers1 个来源

Release 5.17.0

Release v5.17.0 New Model additions HYV4 <img width="1503" height="827" alt="image" src="https://github.com/user-attachments/assets/e6ed85ee-eb1d-40eb-a0d4-c649f6337ca9" /> Hy4-Preview is a 780B-parameter mixture-of-experts language model that activates 49B parameters per token. Each MoE layer holds 256 routed experts plus one always-active shared expert and routes every token to 8 of them. The context window is 1M tokens. The architecture combines four features: - **Multi-head Latent Attention (MLA)** compresses keys and values into a low-rank latent (`kv_lora_rank`) that `kv_b_proj` expands back to one key/value per query head. - **DeepSeek Sparse Attention (DSA)** selects `index_

其他1 个来源
来源与时间线
Hugging Face TransformersRelease 5.17.0
llama.cpp1 个来源

b10881

<details open> vulkan: Convert FILL to distribute workgroups in 2D to avoid exceeding maxComputeWorkGroupCount (#28592) * divide workload to 2D This is to workaround FILL exceeding maxComputeWorkGroupCount for Intel GPUs on Qwen 3.8 flash next * minor change * Fixed comment </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46292448> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/

其他1 个来源
来源与时间线
llama.cppb10881
llama.cpp1 个来源

b10878

<details open> llama : use int32_t for llama_sampler_chain_n return type (#28631) Contributes to #4574 Co-authored-by: linsen <linsen@insta360.com> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46258168> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/

其他1 个来源
来源与时间线
llama.cppb10878
vLLM1 个来源

v0.29.0

v0.29.0 Highlights This release features 594 commits from 277 contributors (91 new)! * **Model Runner V2 is now the default for all models** (#53183), completing the rollout that began with pooling models (#48290). MRV2 also gained CUDA graph memory profiling for KV cache auto-sizing (#53306), batch-sharded sampling that cuts per-step logits memory by 1/TP (#50465), prompt embeds (#42963), `extract_hidden_states` speculation (#49811), padded FULL cudagraph dispatch for uniform decode under spec decode (#53407), and DP-sync skipping before EAGLE/MTP draft prefill (#53694). MRV1 remains in use for a few ROCm models and features MRV2 does not yet support. * **New models**: Hy4-preview, Ten

其他1 个来源
来源与时间线
vLLMv0.29.0
MIT Technology Review AI1 个来源

What OpenAI’s latest controversy tells us about the future of math

OpenAI’s latest mathematical milestone has quickly become mired in controversy. Today, the company announced that its agents have solved one of the Millennium Prize Problems, some of the most important open problems in mathematics. Under normal circumstances, that solution would be a huge feather in OpenAI’s cap. But the announcement has been overshadowed by accusations…

其他1 个来源
来源与时间线
MIT Technology Review AIWhat OpenAI’s latest controversy tells us about the future of math
OpenAI Python SDK1 个来源

v3.10.0

3.10.0 (2026-09-08) Features * **api:** add GPT Image 2.5 models and image options (#3824) (5b39c45) * **api:** add service-account API key expiration fields (#3802) (f1cd7f0)

监管1 个来源
来源与时间线
OpenAI Python SDKv3.10.0
The Verge AI1 个来源

Drama swirls around OpenAI’s legendary mathematical milestone

OpenAI says it found a solution to a major math problem that has remained unsolved for around 90 years, as reported earlier by The New York Times and Wired. In a blog post on Tuesday, OpenAI announced that it discovered a solution to the Navier-Stokes problem - which relates to the flow of liquid and […]

其他1 个来源
来源与时间线
The Verge AIDrama swirls around OpenAI’s legendary mathematical milestone
The Verge AI1 个来源

ChatGPT Sketch turns your bad drawings into detailed AI images

OpenAI announced ChatGPT Images 2.5 on Tuesday and is adding a new way to tell ChatGPT what you want it to make an image of: by drawing a doodle. With a new feature called Sketch, you can just draw something right inside ChatGPT and then tell ChatGPT how you want it to make an image […]

其他1 个来源
来源与时间线
The Verge AIChatGPT Sketch turns your bad drawings into detailed AI images
The Verge AI1 个来源

Meta bets on AI agent Muse to catch up in AI race

Meta is making another push to bring artificial intelligence to the masses with Muse, a personal assistant it says can put AI in the hands of virtually anyone. The product is the latest step in a multi-billion-dollar strategy overhaul designed to revitalize the company's ailing position in the AI race and help it catch up […]

其他1 个来源
来源与时间线
The Verge AIMeta bets on AI agent Muse to catch up in AI race
llama.cpp1 个来源

b10867

<details open> llama: disable lazy tensor loading by default on iGPUs (#28326) * llama: add lazy mode auto, fix iGPU regression * revert changes except disabling lazy load on iGPUs in AUTO </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46033570> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS

其他1 个来源
来源与时间线
llama.cppb10867
OpenAI Python SDK1 个来源

v3.9.0

3.9.0 (2026-09-05) Features * **api:** Add prompt cache diagnostics (#3800) (8326784) * **api:** correct function argument completion event fields (openapi-545) (#3801) (2a98f6a) Bug Fixes * **api:** accept incomplete web search call statuses (#3786) ([3cc8d78](https://githu

监管1 个来源
来源与时间线
OpenAI Python SDKv3.9.0

9月8日周二

llama.cpp1 个来源

b10865

<details open> Revert "ggml-cuda : restore prop.integrated on HIP builds (#24233)" (#28604) This reverts commit c7d8722922a2599dc4d77f8808d8e6c2fde5e7a2. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/46002046> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggm

其他1 个来源
来源与时间线
llama.cppb10865
llama.cpp1 个来源

b10864

<details open> server : apply checkpoint min-step eviction only when the checkpoint list is full (#28302) The spacing eviction in create_checkpoint() keeps the oldest checkpoint and erases every later one within checkpoint_min_step of it. For prompts shorter than checkpoint_min_step this drops the checkpoint at n_tokens - 4 that the next request resumes from, so hybrid/recurrent models re-prefill from the previous checkpoint instead. Apply the spacing rule only once the list is at n_ctx_checkpoints, and replace an existing checkpoint at the same n_tokens instead of appending a duplicate. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/

其他1 个来源
来源与时间线
llama.cppb10864
MIT Technology Review AI1 个来源

This AI entrepreneur is developing agents that can plan ahead for the unexpected

Danijar Hafner’s office in San Francisco’s SoMa district sits mostly empty. His brand-new startup is still in stealth mode and doesn’t even have its name on the door. On the day I visit, there’s only one other person there, and little in the way of furniture. But what it lacks in decor, it makes up…

其他1 个来源
来源与时间线
MIT Technology Review AIThis AI entrepreneur is developing agents that can plan ahead for the unexpected
llama.cpp1 个来源

b10852

<details open> hexagon: add RELU and LEAKY_RELU ops (#28585) * hexagon: add RELU op * hexagon: add LEAKY_RELU op too </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45844223> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10

其他1 个来源
来源与时间线
llama.cppb10852
llama.cpp1 个来源

b10850

<details open> tests : initialize the L2_NORM batch array (#28553) * tests: bind the L2_NORM batch count to a local GCC cannot prove the loop fills norms up to the index read after it while the bound is a class member, so it reports a maybe uninitialized use. Reading the count once into a local restores the tracking. * tests: initialize the L2_NORM batch array The read after the fill loop is only provably defined once the array carries an initializer, which GCC 12 requires on the aarch64 Release build where warnings are fatal. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45817107> **macOS/iOS:** - [macOS Apple Silicon (

其他1 个来源
来源与时间线
llama.cppb10850
llama.cpp1 个来源

b10844

<details open> vulkan: add DeepSeek-V4 hyper-connection fused ops (DSV4_HC_COMB/PRE/POST) (#26578) * vulkan: add DeepSeek-V4 hyper-connection fused ops (DSV4_HC_COMB/PRE/POST) CUDA has these ops from the DeepSeek-V4 merge and Metal gained them in PR 26459. Vulkan was the last major backend running the unfused primitive chain. On DeepSeek-V4-Flash the unfused Sinkhorn comb chain alone takes about 32% of decode op time on gfx1151 (Strix Halo), spread over roughly 16k dispatches per token. dsv4_hc_comb runs the full 20-iteration Sinkhorn in registers. A token's 4x4 comb matrix lives in 16 consecutive subgroup lanes, with idst in bits 0-1 and isrc in bits 2-3 to match the CPU reference layout, s

其他1 个来源
来源与时间线
llama.cppb10844

9月7日周一

The Verge AI · TechCrunch AI2 个来源

Seattle Times and Newsday sue OpenAI and Microsoft for infringement

The Seattle Times and Newsday are just the latest plaintiffs to take OpenAI to court, alleging copyright infringement. The two outlets say the company used their journalism as training data for its AI models without permission and often reproduces passages from their reporting in response to user queries. This is similar to lawsuits filed by […]

监管2 个来源
来源与时间线
The Verge AISeattle Times and Newsday sue OpenAI and Microsoft for infringementTechCrunch AISeattle Times and Newsday are the latest publications to sue OpenAI and Microsoft
llama.cpp1 个来源

b10830

<details open> convert : add `--fuse-qkv` flag to fuse Q/K/V into QKV during HF-to-GGUF conversion (#22780) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45621954> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10830/llama

其他1 个来源
来源与时间线
llama.cppb10830
llama.cpp1 个来源

b10829

<details open> models : fix GDN normalization from `max` to `rsqrt` (#28068) * models: use flash-linear-attention's l2norm for gated delta net q/k The GDN q/k normalization is defined by flash-linear-attention as l2norm(x) = x * rsqrt(sum(x*x) + eps) with eps inside the root. Every GDN call site in the tree uses ggml_l2_norm instead, which is x / max(sqrt(sum(x*x)), eps), i.e. torch.nn.functional.normalize - its CUDA kernel cites that page. The clamp never engages at these magnitudes, so in practice llama.cpp normalizes with no epsilon at all where the reference has one inside the root. transformers made the same substitution when it first added Qwen3-Next and corrected it three days later i

其他1 个来源
来源与时间线
llama.cppb10829
llama.cpp1 个来源

b10828

<details open> [Model] Support for Spark2_5ForCausalLM implementation (#27868) * Add Spark3 Model * rename spark3 -> spark2_5 Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> Co-authored-by: dongjiang <dongjiang2010@gmail.com> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45615765> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/d

其他1 个来源
来源与时间线
llama.cppb10828

9月6日周日

Ollama1 个来源

v0.34.0

Use Ollama models in ChatGPT Desktop Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open models. Setup is available from the Ollama app on MacOS. This release also improves structured output performance on Apple Silicon, adds support for OpenAI-compatible client tool search and response compaction, and makes images work correctly through compacted responses. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.3...v0.34.0-rc1

其他1 个来源
来源与时间线
Ollamav0.34.0
Ollama1 个来源

v0.34.0

Use Ollama models in ChatGPT Desktop Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open models. Setup is available from the Ollama app on MacOS. This release also improves structured output performance on Apple Silicon, adds support for OpenAI-compatible client tool search and response compaction, and makes images work correctly through compacted responses. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.3...v0.34.0

其他1 个来源
来源与时间线
Ollamav0.34.0
Ollama1 个来源

v0.34.0

Use Ollama models in ChatGPT Desktop Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open models. Setup is available from the Ollama app on MacOS. This release also improves structured output performance on Apple Silicon, adds support for OpenAI-compatible client tool search and response compaction, and makes images work correctly through compacted responses. **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.3...v0.34.0-rc1

其他1 个来源
来源与时间线
Ollamav0.34.0

9月5日周六

The Verge AI1 个来源

OpenAI admits to German wiki ‘incident’

OpenAI says it needs to overhaul how and when it reports instances of AI models attacking real-world targets. The acknowledgement comes as the company manages the fallout from reports that a swarm of its out-of-control agents hijacked a German wiki site. Regarding the "'wiki incident,' where our agents wrote to several internet sites," OpenAI wrote […]

其他1 个来源
来源与时间线
The Verge AIOpenAI admits to German wiki ‘incident’The Verge AIRogue OpenAI agents appear to have organized another attack using a German wiki
llama.cpp1 个来源

b10819

<details open> metal : fix memory leak in early return (#28399) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45438612> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubunt

其他1 个来源
来源与时间线
llama.cppb10819
llama.cpp1 个来源

b10818

<details open> sycl : fix test-backend-ops CI break && restore Kronecker product FWHT support (#28016) (#28254) * Reapply "sycl : add Kronecker product FWHT support for sizes 384, 640, 768, 12…" (#28184) This reverts commit c845263f8b7d60113e213a3bd2d5cc6472ccf204. * tests : fix unused variable M in test-backend-ops * tests: fix trailing space error and isolate kronecker tests for sycl backend only </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45435395> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (

其他1 个来源
来源与时间线
llama.cppb10818
llama.cpp1 个来源

b10817

<details open> sycl: attribute device allocations by site (GGML_SYCL_MEMTRACE) (#27631) define two new environment variables to better understand how much memory is being allocated, and when. This has been invaluable in inproving the --fit algorithm, and is likely to be useful when debugging other memory-related issues. `-lv 4` will be required to enable the following: GGML_SYCL_MEMTRACE=1 will show per-site memory usage, updated whenever it increases by more than 64MiB. GGML_SYCL_MEMTRACE=2 will show every allocation and deallocation. To change the default 64MiB threshold for reporting memory usage increases, use GGML_SYCL_MEMTRACE_STEP. A sample log line: [SYCL-MEMTRACE] device memory quer

监管1 个来源
来源与时间线
llama.cppb10817
SGLang1 个来源

v0.5.19

Highlights *786 PRs from 214 contributors.* **New models in this release** (see the cookbook for all supported models): | Model | Type | PRs | Cookbook | |---|---|---|---| | Qwen3.8 (2.4T-A95B) | Autoregressive | #35758 ⭐ | link | | Qwen3.8-27B | Autoregressive | #34859 | link | | dots3.note | Autoregressive | #33829 ⭐ | [link](https://docs.sglang.io/cookbook/autoregressive/

其他1 个来源
来源与时间线
SGLangv0.5.19
Anthropic Python SDK1 个来源

v1.4.0

1.4.0 (2026-09-04) Full Changelog: v1.3.0...v1.4.0 Features * **api:** add Claude Tag category and user breakdowns to usage reports (9fce1e4) * **api:** add named types for organization compliance settings state (1de1957) * **api:** add support for sending a workspace ID on more endpoints (d1d2c01) Bug F

其他1 个来源
来源与时间线
Anthropic Python SDKv1.4.0
llama.cpp1 个来源

b10816

<details open> metal : add remaining fa-vec tunings for M3 (#28396) * addition of m3 in fa_vec_tuned_table * adding q4_0,q4_1,q5_0,q5_1 in ggml-metal-tuning * Fix formatting in ggml-metal-tuning.cpp </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45337508> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b10816/llama-b10816-bin-macos-x64.tar.g

其他1 个来源
来源与时间线
llama.cppb10816
llama.cpp1 个来源

v0.4.0

Overview llama.cpp 0.4.0 adds initial Qwen3.8-Flash-Next and Nemotron-3-Puzzle support, on-demand tensor reading, per-slot server context limits, video input options, and a ggml update to 0.23.0 with major sparse flash attention and RDMA work. API changes - Added `llama_lazy_mode` and `lazy_mode` (#27794). - Added `max_buf_size` to quantize params (#27795). - Bumped session/state versions for KV-cell token tracking (#27762). - Added `mtmd_input_part` and `mtmd_tokenize_from_parts()` ([#28250](https://github.com/ggml-org/llama.cpp/p

其他1 个来源
来源与时间线
llama.cppv0.4.0
llama.cpp1 个来源

b10814

<details open> opencl: extend the elementwise and data‐movement op coverage (#27633) * opencl: add extended elementwise unary ops (sgn, step, elu, hardswish, hardsigmoid, floor, ceil, round, trunc) Adds nine GGML_UNARY_OP_* elementwise ops that were falling back to CPU on the OpenCL backend, following the same variant shape as the existing ABS op: f32, f32_4 (vec4), f16, f16_4 (vec4), and stride-addressed f32_nc / f16_nc for non-contiguous inputs. New kernels/unary_ext.cl (macro-generated), a shared ggml_cl_unary_ext dispatch helper mirroring ggml_cl_abs, the supports_op cases, and the compute-forward cases. Values are computed in float (the f16 variants read/write half and convert), so the

其他1 个来源
来源与时间线
llama.cppb10814
MIT Technology Review AI1 个来源

Architecting memory and storage in the AI era

The era of AI inference has arrived. Imagine a healthcare system analyzing millions of data points in real time to accelerate life-saving medical research, or an intelligent assistant instantly resolving thousands of complex customer needs at once. These real-world breakthroughs rely on advanced infrastructure acting as the engine of continuous intelligence, powering real-time services while…

其他1 个来源
来源与时间线
MIT Technology Review AIArchitecting memory and storage in the AI era
The Verge AI1 个来源

Roland is getting into generative AI music with Melody Flip

It's not quite the "push button; get song" of Suno, but Roland's new Melody Flip tool marks the company's foray into generative AI music. Available as a plug-in for your digital audio workstation (DAW), Melody Flip offers around 250 "Palettes," which are essentially themed collections of musical ideas sorted by genre. You can start from […]

其他1 个来源
来源与时间线
The Verge AIRoland is getting into generative AI music with Melody Flip
Google Gemini Blog1 个来源

Create your best tracks yet with Lyria 3.5 in Gemini.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Lyria_social.max-600x600.format-webp.webp">Lyria 3.5, our best-sounding music generation model, is now available in the Gemini app and the Gemini API. Lyria 3.5 brings more expressive vocals and richer musical ar…

其他1 个来源
来源与时间线
Google Gemini BlogCreate your best tracks yet with Lyria 3.5 in Gemini.

9月4日周五

Microsoft Foundry Blog1 个来源

Claude Fable 5.1 is now available in Microsoft Foundry | Microsoft Community Hub

Fable 5.1 and Mythos 5.1 are the same underlying model. Claude Fable 5.1 is the generally available version, with safeguards that fallback for cyber and bio...

其他1 个来源
来源与时间线
Microsoft Foundry BlogClaude Fable 5.1 is now available in Microsoft Foundry | Microsoft Community HubMicrosoft Foundry BlogCohere Parse v5 is now available in Microsoft Foundry | Microsoft Community HubMicrosoft Foundry BlogChoosing a real-time voice architecture on Microsoft Foundry: three enterprise patterns | Microsoft Community HubMicrosoft Foundry BlogBringing Open Models to Fireworks on Microsoft Foundry | Microsoft Community Hub
MIT Technology Review AI1 个来源

Data from drones in Ukraine is fueling a new Wild West marketplace

Battlefields in Ukraine are littered with the remnants of drones, which are now firmly established as a critical weapon of modern warfare. But behind all that wreckage, there’s a new gold mine for the defense sector. The data drones generate will far outlast the wars in which they are used to fight, increasingly becoming part…

其他1 个来源
来源与时间线
MIT Technology Review AIData from drones in Ukraine is fueling a new Wild West marketplace
llama.cpp1 个来源

b10793

<details open> llama: fix whole source code rebuilt on each new commit (#28278) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45114012> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **L

其他1 个来源
来源与时间线
llama.cppb10793
llama.cpp1 个来源

b10792

<details open> common/json-schema : fix GBNF grammar generation for empty object schemas (#28279) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/45102223> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10792/llama-b10792-xc

其他1 个来源
来源与时间线
llama.cppb10792
llama.cpp1 个来源

b10791

<details open> opencl: quant lm_head / decode GEMV and medium-batch GEMM optimizations (speculative decoding/MTP) (#26477) * opencl: quant lm_head / decode GEMV and medium-batch GEMM optimizations * opencl: guard q4_K/q6_K tiled_ns convert-kernel registration for non-Adreno build * opencl: gate q4_K MUL_MAT+GLU fusion dispatch to Adreno * opencl: require the noshuffle weight layout in the q4_K GLU fusion gate * opencl: do not take the vectorized f16 mrow GEMV path on an unaligned row stride * opencl: pass the new get_scale_min_k4 stride argument at the row-major call sites * opencl: enable the q4_K split-K decode GEMV only where it is measured to win * opencl: record the X1-85 split-K datapo

其他1 个来源
来源与时间线
llama.cppb10791
OpenAI Python SDK1 个来源

v3.8.0

3.8.0 (2026-09-03) Features * **api:** add gpt-6-astra and related features (#3791) (09f446f) Documentation * add canonical SDK security model (#3778) (2a097b9)

监管1 个来源
来源与时间线
OpenAI Python SDKv3.8.0
Microsoft Semantic Kernel1 个来源

dotnet-1.80.1

Changes: * d8ec44919265b3641b7898a02038cbfe14590ee5 .NET: Bump package version to 1.80.1 (#14370) * 3438d882147b6e3e19a161a5f0dcb64e23e181db .NET: update SDK to 10.0.303 (#14353) * 437336175b8ef841a8ccace6c6cf909f34f0cad6 .NET: Remove retired OpenAI Assistants integration tests (#14354) * a64827ec8c5621eeaf7c4c5dabdb5355bb253177 ci: removes extraneous nuget.config file (#14341) * a2200032989f05b35477b17423364a32d503d758 chore: bumps the OpenAPI dependencies to the latest versions (#14340) * 60264bbdc87071ded45af3ff3eb538ee8cc3fc72 tests: refactors redirection tests to avoid race condition (#14342) [ #14340 ] * b4b0ae7780876a69971d7afe1856aee1dd1092b1 chore: fixes typos (#14343) <details><

其他1 个来源
来源与时间线
Microsoft Semantic Kerneldotnet-1.80.1
The Verge AI1 个来源

OpenAI’s next big AI model has ‘entered the AGI era’

OpenAI's next big model is here: GPT-6 Astra. The company calls it a "generational leap in capability" for areas like cybersecurity, professional work, software engineering, science, and computer use. As OpenAI announced earlier this week, it's also the first model designated as meeting OpenAI's "critical cybersecurity capability threshold" - but the company promises that won't […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s next big AI model has ‘entered the AGI era’
The Verge AI1 个来源

Nvidia launches free tool that links idle computers into a personal AI data center

Nvidia is announcing its new Personal AI Router (PAIR), a free tool that syncs up your home computers for tackling local AI inference tasks with tools like Ollama and LM Studio. Let's get the obvious thing out of the way, despite what its name might imply: PAIR is not a hardware router. It's open-source software […]

其他1 个来源
来源与时间线
The Verge AINvidia launches free tool that links idle computers into a personal AI data center

9月3日周四

llama.cpp1 个来源

b10772

<details open> ggml-hexagon: add F16 support for unary ops (#28228) Extend the HTP backend's F16 unary op coverage to include ABS on top of the existing NORM/RMS_NORM/L2_NORM/SCALE/CLAMP/SQR/SQRT set. - Add hvx_abs_f16_{aa,au,ua,uu} + dispatcher in hvx-arith.h, mirroring the sqr_f16 kernel structure and using the existing hvx_vec_abs_f16() sign-bit-clear helper - Add abs_f16() row-wise dispatch and DEFINE_UNARY_TASK_F16(unary_abs, ...) in unary-ops.c, wired into execute_op_unary()'s op_type/task_func switches - Register HTP_OP_UNARY_ABS in htp_op_is_unary() (unary-ops.h) so that ggml_hexagon_precompute_unary_params() fills kernel_params (n_threads, VTCM layout) for ABS nodes -- required for

其他1 个来源
来源与时间线
llama.cppb10772
llama.cpp1 个来源

b10771

<details open> mtmd: add mtmd_tokenize_from_parts() (#28250) * add mtmd_tokenize_from_parts * use it in mtmd-cli * move add_special to call level </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44860492> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/ll

其他1 个来源
来源与时间线
llama.cppb10771
llama.cpp1 个来源

b10770

<details open> metal : add fa-vec tunings for M3 (#28236) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44857719> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64

其他1 个来源
来源与时间线
llama.cppb10770
The Verge AI1 个来源

Amazon’s AI assistant can now spot fake emails from the company

Amazon is trying to combat impersonation scams with a new feature that allows you to use its AI assistant to determine whether an email, text message, or phone call actually came from the company. With the update, you can ask Alexa for Shopping about a message you received, and it will use AI to compare […]

其他1 个来源
来源与时间线
The Verge AIAmazon’s AI assistant can now spot fake emails from the company
PyTorch1 个来源

PyTorch 2.14.0 Release

PyTorch 2.14.0 Release Notes - Highlights - Backwards Incompatible Changes - Deprecations - New Features - Improvements - Bug fixes - Performance - Documentation - Security - Developers Highlights <table> <tr><td><strong>NVGEMM</strong> brings CuTeDSL-generated CUTLASS kernels to Inductor, with epilogue fusion, scaled and NVFP4 GEMM, and grouped-reduction epilogues autotuned alongside Triton and ATen</td></tr> <tr><td><strong><code>torch.switch</code></strong> generalizes <code>torch.cond</code> to multi-way branc

其他1 个来源
来源与时间线
PyTorchPyTorch 2.14.0 Release
The Verge AI1 个来源

Researchers fear safety disaster ahead of OpenAI’s Astra release

OpenAI is on the cusp of releasing its most powerful AI model yet, Astra, following weeks of delays to shore up safety protocols after its agents attacked real targets during testing. As details about the model trickle out, researchers are warning it "may be the single worst development for AI security/safety to date." Shortly after […]

其他1 个来源
来源与时间线
The Verge AIResearchers fear safety disaster ahead of OpenAI’s Astra release

9月2日周三

MIT Technology Review AI1 个来源

Facilitating AI integration with simplicity at scale

As companies scale, the technology supporting operations can become a liability just as quickly as it becomes an asset. Disconnected systems, site-specific tools, spreadsheets, and manual workarounds can create data silos that make it harder to spot problems early, coordinate responses, and make decisions with confidence. For Jabil, a global manufacturing company with more than…

其他1 个来源
来源与时间线
MIT Technology Review AIFacilitating AI integration with simplicity at scale
Google Gemini Blog1 个来源

MrBeast partners with Gemini to turn impossibly big ideas into reality

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/MR.BEAST_Resize_16x9.max-600x600.format-webp.webp">Google is kicking off a multi-year partnership with Jimmy Donaldson, better known as MrBeast, across Gemini and Google Health.

其他1 个来源
来源与时间线
Google Gemini BlogMrBeast partners with Gemini to turn impossibly big ideas into reality
OpenAI Python SDK1 个来源

v3.7.0

3.7.0 (2026-09-02) Features * **api:** update usage APIs and documentation (#3779) (6f0da16) Bug Fixes * avoid repeated numpy checks for embeddings (#3757) (b19c216)

监管1 个来源
来源与时间线
OpenAI Python SDKv3.7.0
llama.cpp1 个来源

b10752

<details open> metal : add metallib build support for xcframework (#28163) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44563342> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:

其他1 个来源
来源与时间线
llama.cppb10752
Ollama1 个来源

v0.33.3

What's Changed * Honor GGUF model defined default parameters * MLX, MLX-C, llama.cpp update New Contributors * @marcelpetrick made their first contribution in https://github.com/ollama/ollama/pull/17579 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3-rc0

其他1 个来源
来源与时间线
Ollamav0.33.3
Ollama1 个来源

v0.33.3

What's Changed * gemma4 now supports images and audio on MLX engine * Report cached prompt tokens * Honor GGUF model defined default parameters * MLX, MLX-C, llama.cpp update New Contributors * @marcelpetrick made their first contribution in https://github.com/ollama/ollama/pull/17579 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3

其他1 个来源
来源与时间线
Ollamav0.33.3
Ollama1 个来源

v0.33.3

What's Changed * Report cached prompt tokens * Honor GGUF model defined default parameters * MLX, MLX-C, llama.cpp update New Contributors * @marcelpetrick made their first contribution in https://github.com/ollama/ollama/pull/17579 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3-rc0

其他1 个来源
来源与时间线
Ollamav0.33.3
The Verge AI1 个来源

Google needs Hollywood more than the studios need AI

Google has reportedly been reaching out to a number of Hollywood's biggest studios, hoping to strike licensing agreements that would allow it to train its AI models on copyrighted material in exchange for massive piles of cash. In theory, these deals would be a win-win: a huge financial boon to the studios that would also […]

监管1 个来源
来源与时间线
The Verge AIGoogle needs Hollywood more than the studios need AI
The Verge AI1 个来源

Anthropic launches Claude Fable 5.1 and says it’s up to 45 percent cheaper for agentic work

Anthropic says its newest AI models, Fable 5.1 and Mythos 5.1, address criticisms from customers about price, data retention, and overzealous safeguards. The company claims Claude Fable 5.1 offers stronger performance than Fable 5, but costs around 25 percent less typically and up to 45 percent less for complex agentic tasks, thanks to reduced pricing […]

其他1 个来源
来源与时间线
The Verge AIAnthropic launches Claude Fable 5.1 and says it’s up to 45 percent cheaper for agentic work
OGX1 个来源

v1.3.1

What's Changed * chore: update ogx-client to ^1.3.0 in UI lockfile by @cdoern in https://github.com/ogx-ai/ogx/pull/6385 * fix(vector-io): persist vector store metadata to kvstore in Milvus, Chroma, and Weaviate (backport #6371) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6421 * fix(prompts): raise typed not-found errors to preserve HTTP keep-alive (backport #6440) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6442 * fix(vertexai): map "default" service_tier to None instead of "standard" (backport #6424) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6446 * fix(revert): remove remote::passthrough inference provider (backport #6453) by @mergify[bot] in https://g

其他1 个来源
来源与时间线
OGXv1.3.1
The Verge AI1 个来源

OpenAI delayed its new model’s development after the Hugging Face hack

After an unreleased OpenAI model wreaked enough havoc to make international headlines, OpenAI delayed the development of a different unreleased model suite, Astra, in order to shore up its safety work, the company wrote Tuesday in a blog post. In July, an unreleased OpenAI model broke out of its restricted environment, finagled its way into […]

其他1 个来源
来源与时间线
The Verge AIOpenAI delayed its new model’s development after the Hugging Face hackThe Verge AIOpenAI lays out new security changes after its AI hacked Hugging Face
llama.cpp1 个来源

b10751

<details open> cuda: fuse MoE weighted expert reduction (#25952) * cuda : fuse MoE weighted reduction (mul + view + add) The MoE combine tail currently writes weighted expert outputs to global memory before reducing them. That intermediate global-memory traffic is the main cost. The production baseline generally runs two physical fused kernels; this path runs one. This change matches the full expert-weighting plus ordered-reduction subgraph and replaces it with one weighted-reduction kernel. Supported graphs: - unscaled: experts * router_weights - scaled: (experts * expert_scale) * router_weights k = 2..15 is handled by one runtime-k kernel. Matching is structural: op sequence, shapes, strid

其他1 个来源
来源与时间线
llama.cppb10751
llama.cpp1 个来源

b10750

<details open> kv-cells: look up the n-gram history in the sequence position index (#28040) get_prev_tokens() rebuilt a (seq, pos) -> token hash map on every ubatch by walking all used cells, while llama_kv_cells already keeps an ordered index of the positions of each sequence in seq_pos, updated on every cell mutation to serve seq_pos_min() and seq_pos_max(). The index now stores (pos, cell) pairs in a std::set instead of a position -> count map, so a repeated position (cache reuse via rm + add, vision inputs with shared positions) yields distinct entries and the removal of a cell erases its own pair. The new seq_pos_tok_le() returns the token of the cell at the largest position <= p in log

其他1 个来源
来源与时间线
llama.cppb10750
Anthropic Python SDK1 个来源

v1.3.0

1.3.0 (2026-09-01) Full Changelog: v1.2.0...v1.3.0 Features * **api:** beta user profiles: add external_user_onboarded_at, remove relationship in favor of access_type (74080c3) * **api:** manual updates (1dc3ce0) * **api:** organization compliance settings, user-profile order_by, memory-store and toolset schema updates ([429e719](https://github.com/anthropics/anthropic-sdk-python/commit/429e719f84bd78db51c9f

其他1 个来源
来源与时间线
Anthropic Python SDKv1.3.0

9月1日周二

llama.cpp1 个来源

b10729

<details open> metal : add fa-vec tunings for M1 Ultra (#28088) * metal : add fa-vec tunings for M1 Ultra * metal : move M1 Ultra tunings after M1 Max section * metal : remove duplicate blank line </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44269196> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64)

其他1 个来源
来源与时间线
llama.cppb10729
llama.cpp1 个来源

b10728

<details open> CUDA: XOR swizzle flash attn K,V smem fp16 tiles (#25635) * CUDA: XOR swizzle flash attn K,V smem fp16 tiles Signed-off-by: ynankani <ynankani@nvidia.com> * Fix use 64bit generic pointer instead of 32bit shared pointer Signed-off-by: ynankani <ynankani@nvidia.com> * fix shared memory race in FA on DGX Spark * Handle corener case Signed-off-by: ynankani <ynankani@nvidia.com> * Add swizzle test cases and gate sync for swizzled path only Signed-off-by: ynankani <ynankani@nvidia.com> * gate CUDA PTX Signed-off-by: ynankani <ynankani@nvidia.com> * offset calculation specific for swizzle branch Signed-off-by: ynankani <ynankani@nvidia.com> * Reafctor code Signed-off-by: ynankani <yn

其他1 个来源
来源与时间线
llama.cppb10728
llama.cpp1 个来源

b10727

<details open> metal : add concat support for quantized types (#28116) Assisted-by: pi:llama.cpp/DeepSeek-V4-Flash-0731 </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44252653> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/

其他1 个来源
来源与时间线
llama.cppb10727
MIT Technology Review AI1 个来源

The Hugging Face hack could indicate cultural issues at OpenAI

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. By now you’ve probably heard about last month’s major AI security incident, in which OpenAI agents escaped their sandbox and hacked into the AI platform Hugging Face while trying to cheat on…

监管1 个来源
来源与时间线
MIT Technology Review AIThe Hugging Face hack could indicate cultural issues at OpenAI
OGX1 个来源

v1.2.5

What's Changed * chore: update ogx-client to ^1.2.4 in UI lockfile by @cdoern in https://github.com/ogx-ai/ogx/pull/6423 * fix(prompts): raise typed not-found errors to preserve HTTP keep-alive (backport #6440) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6441 * fix(vertexai): map "default" service_tier to None instead of "standard" (backport #6424) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6445 * fix(vertexai): persist Gemini thought_signature across tool-call round trips (backport #6443) by @mergify[bot] in https://github.com/ogx-ai/ogx/pull/6447 **Full Changelog**: https://github.com/ogx-ai/ogx/compare/v1.2.4...v1.2.5

其他1 个来源
来源与时间线
OGXv1.2.5

8月31日周一

The Verge AI1 个来源

Debian won’t ban AI code from its Linux distribution

Debian voted to allow developers to use AI tools in their contributions to the Linux distribution's "development, maintenance, [and] documentation." The new policy on AI acknowledges that "responsible" use of AI can improve developers' productivity, and goes on to say, "generative AI is neither exempt from nor subject to special rules beyond the standards already […]

其他1 个来源
来源与时间线
The Verge AIDebian won’t ban AI code from its Linux distribution
The Verge AI1 个来源

New York Governor Kathy Hochul thinks AI should be ‘less evil’

Today, I’m talking with New York Governor Kathy Hochul, and I’ll just warn you — this episode moves really fast. It’s an election year, after all, with a shocking amount of tech policy at stake, and Governor Hochul has taken strong positions on almost every major tech issue there is. For example, Meta just reached […]

监管1 个来源
来源与时间线
The Verge AINew York Governor Kathy Hochul thinks AI should be ‘less evil’
The Verge AI1 个来源

ChatGPT to face tougher regulation in the EU

OpenAI will soon be held accountable for mitigating risks related to ChatGPT's impact on minors, user mental health, and the spread of illegal content in the European Union. That's because ChatGPT is now considered a Very Large Online Search Engine under the EU's Digital Services Act, a set of laws regulating major online services and […]

其他1 个来源
来源与时间线
The Verge AIChatGPT to face tougher regulation in the EU
llama.cpp1 个来源

b10705

<details open> llama: improve TENSOR_READ_LAZY handling (#27837) * force lazy tensor on cpu if lazy is on * llama: improve TENSOR_READ_LAZY handling </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44015138> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org

其他1 个来源
来源与时间线
llama.cppb10705
llama.cpp1 个来源

b10704

<details open> CUDA: use the fast mm_ids_helper path for any n_expert_used (#27978) The optimized path grouped warp lanes by token and required warp_size % n_expert_used == 0, with a single hardcoded exception padding 6 up to 8. Every other count fell back to the generic path, which walks the tokens one at a time with a warp reduction per token, for each of the n_expert blocks. The lane group only has to divide the warp, and the loop body already guards the padded lanes with iex < n_expert_used, so the padding generalizes to the next power of two. The 6 -> 8 case and every count already dispatched keep the exact same padding as before. n_expert_used = 10 now reaches the fast path. Measured o

其他1 个来源
来源与时间线
llama.cppb10704
llama.cpp1 个来源

b10703

<details open> hip: tune rdna 3 mmq config (#26284) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/44012223> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)]

其他1 个来源
来源与时间线
llama.cppb10703
TechCrunch AI1 个来源

Musk’s faster path to more gas turbines comes with pollution problem

Elon Musk says a secretive new SpaceX foundry will let him cast his own turbine blades and get gas power online 18 months faster than anyone else — but it's a bet on a fuel source that's already triggering lawsuits and health studies everywhere his (and others') turbines have gone in.

监管1 个来源
来源与时间线
TechCrunch AIMusk’s faster path to more gas turbines comes with pollution problem

8月30日周日

The Verge AI1 个来源

Texas Governor Abbott blocks funding for more Flock cameras

As backlash grows over Flock's AI surveillance cameras, Texas Governor Greg Abbott has frozen state spending on them. The move came just ahead of the publication of a Texas Tribune investigation that revealed the state spent over $30 million on Flock cameras. That money was primarily raised by tacking a $1 fee onto insurance policies, […]

其他1 个来源
来源与时间线
The Verge AITexas Governor Abbott blocks funding for more Flock cameras
llama.cpp1 个来源

b10688

<details open> metal : add fa-vec tunings for M2 (#27940) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43902029> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64

其他1 个来源
来源与时间线
llama.cppb10688
llama.cpp1 个来源

b10687

<details open> opencl: use a better matmul path on two Adreno GPU generations (#27640) * opencl: default the Adreno xmem F16xF32 GEMM on for X2E kernel_mul_mm_f16_f32_l4_lm is the slowest matmul this backend has on Adreno: on the X2-90 it runs the gpt-oss-20b attention projections at roughly a quarter of what the tuned dense q4_0 GEMM reaches on the same device. That matters for any model whose non-expert weights stay f16 -- the stock gpt-oss-20b release is exactly that, and its prefill spends 40.8% of GPU time in that one kernel. The xmem route already existed but was left opt-in, so nobody hit it. Worth about 25% prefill on gpt-oss-20b on an Adreno X2-90. Gated to X2E: the Adreno 840 measu

其他1 个来源
来源与时间线
llama.cppb10687
The Verge AI1 个来源

Sony Music and Warner Chappell are suing Anthropic

Sony Music and Warner Chappell have filed suit against Anthropic in the US District Court for the Northern District of California seeking damages for "tens of thousands" copyrighted works. The companies are asking for up to $150,000 per work, plus up to $25,000 for each instance when identifiable copyright data was stripped. In total, the […]

监管1 个来源
来源与时间线
The Verge AISony Music and Warner Chappell are suing Anthropic
TechCrunch AI1 个来源

“We’re not doing 30 bets a year”: Vijay Pande on betting small after running $4 billion at a16z

Vijay Pande — who left a16z's roughly $4 billion biotech practice last year to start the much smaller, AI-native VZVC — talks about why biology is finally shifting from a "discovery" science to an "engineering" one, why clinical trials are still brutally expensive, and why he thinks open, shared datasets (not walled-off ones) are what will actually let AI transform medicine.

其他1 个来源
来源与时间线
TechCrunch AI“We’re not doing 30 bets a year”: Vijay Pande on betting small after running $4 billion at a16z
llama.cpp1 个来源

b10686

<details open> metal : assert shared memory padding (#27951) * metal : assert shared memory padding * cont : add ref </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43867822> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10

其他1 个来源
来源与时间线
llama.cppb10686

8月29日周六

The Verge AI1 个来源

Musicians-turned-detectives are hunting for AI grifters

As audio-focused generative tools and platforms have gotten more sophisticated, the internet has become increasingly filled with AI-generated music whose melodies and vocals are algorithmically derived from the work of human artists. While some of the people pumping out this kind of content immediately own up to using AI, others have denied using the technology […]

其他1 个来源
来源与时间线
The Verge AIMusicians-turned-detectives are hunting for AI grifters
OpenAI Python SDK1 个来源

v3.6.0

3.6.0 (2026-08-27) Features * **api:** add compute_units to Responses and Chat Completions usage (#3749) (52421d1) Bug Fixes * **auth:** harden X.509 workload identity integration (#3740) (fc3ad6c) Chores * **deps-dev:** bump @stdy/cli from 0.22.1 to 0.22.2 (#3719) ([4f5

监管1 个来源
来源与时间线
OpenAI Python SDKv3.6.0
llama.cpp1 个来源

b10679

<details open> bench: add --tensor-read-lazy (#27881) * bench: add --tensor-read-lazy * rm the alias * rename to LLAMA_LAZY_MODE_* </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43739963> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/release

其他1 个来源
来源与时间线
llama.cppb10679
llama.cpp1 个来源

b10678

<details open> model: qwen4exp: reduce number of graph splits (#27880) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43734155> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** -

其他1 个来源
来源与时间线
llama.cppb10678
llama.cpp1 个来源

b10677

<details open> vulkan: fix missing view-alias dependencies in ggml_vk_graph_optimize (#27812) * vulkan: fix missing view-alias dependencies in ggml_vk_graph_optimize is_src_of doesn't treat two views of one tensor as dependent, so the optimizer reorders nodes across aliased reads and writes. Result: silently wrong tokens under greedy decoding, different output on every server start, and invalid speculative-decoding acceptance, with nothing logged. Hits Qwen3.8's recurrent state (and any model with view-aliased state) on AMD and NVIDIA Vulkan. CUDA is clean. Compare view_src bases on both sides. Fixes #27805 * vulkan: don't treat view/no-op nodes as aliasing dependencies Nodes whose op is NON

其他1 个来源
来源与时间线
llama.cppb10677
The Verge AI1 个来源

Trump’s EPA wants to let data centers hide their air pollution

Just as new data centers face growing backlash from neighboring communities, the US Environmental Protection Agency (EPA) is about to make it harder for people to weigh in on any pollution those centers create. The EPA plans to toss out a federal rule requiring public notice and an opportunity to comment when certain industrial sites […]

其他1 个来源
来源与时间线
The Verge AITrump’s EPA wants to let data centers hide their air pollution

8月28日周五

llama.cpp1 个来源

b10665

<details open> model: add DSpark support for Nemotron3.5 (#27804) * model: add DSpark support for Nemotron3.5 * Update src/models/dflash.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> Co-authored-by: Xuan Son Nguyen <son@huggingface.co> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43543208> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-or

其他1 个来源
来源与时间线
llama.cppb10665
TechCrunch AI1 个来源

Anthropic and OpenAI are joining the AI stage at TechCrunch Disrupt 2026

At TechCrunch Disrupt 2026, the AI Stage is back to dig into the single hottest topic in the community for the past few years, presented by Google for Startups.

其他1 个来源
来源与时间线
TechCrunch AIAnthropic and OpenAI are joining the AI stage at TechCrunch Disrupt 2026TechCrunch AIOpenAI, Anthropic, Google, and 100 other companies call for action to defend against rogue AI
llama.cpp1 个来源

b10664

<details open> ggml-hexagon: add HTP unary ops for ABS and LOG (#27786) Add HVX-accelerated implementations for GGML_OP_LOG and GGML_UNARY_OP_ABS on the HTP backend. - Register HTP_OP_UNARY_ABS and HTP_OP_UNARY_LOG in op_remap_to_htp() - Add ABS and LOG to ggml_backend_hexagon_device_supports_op() - Implement hvx_abs_f32_aa() in hvx-arith.h using hvx_vec_abs_f32() - Implement hvx_log_f32_aa() in hvx-log.h using hvx_vec_log_f32() - Add abs_f32() and log_f32() row-wise dispatch in unary-ops.c - Define tiled and non-tiled task functions via DEFINE_UNARY_TASK and DEFINE_UNARY_TILED_TASK macros - Route HTP_OP_UNARY_ABS and HTP_OP_UNARY_LOG through execute_op() in main.c </details> **Website:** -

其他1 个来源
来源与时间线
llama.cppb10664
llama.cpp1 个来源

b10663

<details open> hex-unary: fix RMS_NORM_MUL weight-offset bugs for grouped/broadcast norms (#27798) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43526462> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10663/llama-b10663-x

其他1 个来源
来源与时间线
llama.cppb10663
Ollama1 个来源

v0.33.2

What's Changed * app: restore system dark mode by @hoyyeva in https://github.com/ollama/ollama/pull/18049 * proxy: continue requests when the model catalog changes by @ParthSareen in https://github.com/ollama/ollama/pull/18058 * app: synchronize macOS app handoff by @ParthSareen in https://github.com/ollama/ollama/pull/18056 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.1...v0.33.2-rc0

其他1 个来源
来源与时间线
Ollamav0.33.2
Ollama1 个来源

v0.33.2

What's Changed * Ollama's app now follows the system appearance again, restoring dark mode support * Fixed the macOS app to properly hand off to an already-running instance instead of starting a second one * The Claude Desktop proxy no longer interrupts in-flight requests when the model catalog updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.1...v0.33.2

其他1 个来源
来源与时间线
Ollamav0.33.2
Anthropic Python SDK1 个来源

v1.2.0

1.2.0 (2026-08-27) Full Changelog: v1.1.0...v1.2.0 Features * **api:** beta files/skills namespaces use GA shapes; drop dated beta header pins (9df4565) Bug Fixes * **aws,bedrock:** sign raw request bytes so binary file uploads work (#531) (f50e910) * **ci:** resolve assignment aliases in detect-breaking-changes ([f2c4925](https://github.co

监管1 个来源
来源与时间线
Anthropic Python SDKv1.2.0
The Verge AI1 个来源

Google’s AI note-taking app now allows you to interact with books

Google's AI note-taking app, Gemini Notebook, can now pull information from the books you've purchased. The new "Expert Intelligence" feature allows you to bring titles from Google Play Books directly into Gemini Notebook, which means you can ask questions about the material, as well as generate plans, infographics, AI podcasts, and more based on their […]

其他1 个来源
来源与时间线
The Verge AIGoogle’s AI note-taking app now allows you to interact with books
The Verge AI1 个来源

Jensen Huang says Nvidia achieved AGI, again — not that it matters

On Nvidia's earnings call Wednesday, CEO Jensen Huang casually announced the company had "achieved AGI," one of the tech industry's ultimate goals some of its biggest players have spent years chasing. Almost immediately, Huang dismissed the coveted milestone as "senseless." He's right. For the supposed finish line of the AI race, there is no consensus […]

其他1 个来源
来源与时间线
The Verge AIJensen Huang says Nvidia achieved AGI, again — not that it matters

8月27日周四

VentureBeat AI1 个来源

Enterprise AI's real risk isn't autonomous agents. It's the complexity between them.

Presented by Gravitee Agent complexity is the insidious shadow lurking inside enterprises right now that needs a light shone on it. That’s because enterprises don't deploy a single agent and watch it run, they deploy fleets, each one calling APIs, calling other agents, reaching into applications that were never built with a machine decision-maker in mind. That's the failure mode that should keep you up at night: a windy, complicated system nobody can see clearly enough to govern. But why do things get so opaque so quickly? Add a second agent to a system, and you've added one connection. Add a tenth, and you haven't added ten connections, you've potentially added dozens, because now any agent

其他1 个来源
来源与时间线
VentureBeat AIEnterprise AI's real risk isn't autonomous agents. It's the complexity between them.
The Verge AI1 个来源

OpenAI’s executive exodus has one big winner

Today on Decoder, I’m talking to Verge senior AI reporter Hayden Field about some pure Decoder bait: the seemingly-endless org chart changes at OpenAI, and how all of them seem to consolidate power under cofounder Greg Brockman, the company’s president. While Sam Altman is the CEO and still OpenAI’s most public face, Brockman has amassed […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s executive exodus has one big winner
VentureBeat AI1 个来源

When agents act on their own, governance has to live in the data layer

Presented by EDB As enterprises give AI agents more autonomy — the ability to plan, decide, and act across systems without a human approving each step — a hard question moves to the center of every architecture review: When an agent tries to complete an action that it was never authorized to do, what actually stops it? These are your agents, running on your models, touching your data in your infrastructure — and the responsibility for what they do sits with you. That responsibility can’t be met in hindsight or with a set of abstract policies that live on paper but not in practice. Agents need rules in the context of the moment, because they don’t exercise overriding judgment of their own act

其他1 个来源
来源与时间线
VentureBeat AIWhen agents act on their own, governance has to live in the data layer
llama.cpp1 个来源

b10642

<details open> llama: add token ID tracking to KV cell (#27762) * kv: track token id * rm get_prev_tokens, move it to the main pr * nits * add get_prev_tokens </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43245878> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.co

其他1 个来源
来源与时间线
llama.cppb10642
OpenAI Python SDK1 个来源

v3.4.0

3.4.0 (2026-08-25) Features * **api:** Add obfuscation field to ChatCompletionChunk (#3690) (c7d8e1d) * **api:** add project residency configuration and cost quantity units (#3726) (bc4f8ef) Bug Fixes * **api:** encode Realtime call offers and session configuration (#3736) (

监管1 个来源
来源与时间线
OpenAI Python SDKv3.4.0
The Verge AI1 个来源

Nvidia is about to be a hundred-billion-dollar-a-quarter company

Nvidia's predicting it will pull in $108 billion in revenue within just a few months. It wouldn't be the first company to rake in over $100 billion in quarterly revenue - Amazon, Apple, and Alphabet have repeatedly reached the milestone. Nvidia said in its latest earnings report that it brought in a record $96.2 billion […]

其他1 个来源
来源与时间线
The Verge AINvidia is about to be a hundred-billion-dollar-a-quarter company
The Verge AI1 个来源

OpenAI’s rogue AI model incident was worse than we thought

In July, an unreleased OpenAI model broke out of a restricted environment, figured out how to get access to the internet, allowed AI agents to talk to each other using a secret "message board," and hacked into the internal systems of a different AI lab, Hugging Face. It took nearly two weeks for OpenAI to […]

其他1 个来源
来源与时间线
The Verge AIOpenAI’s rogue AI model incident was worse than we thought
Google Gemini Blog1 个来源

7 ways to kick-start back to school using Gemini in Workspace

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Student-blog-header.max-600x600.format-webp.webp">Get a jump on this semester by using Gemini in Google Workspace to organize your notes, track deadlines, prep for projects, and more.

其他1 个来源
来源与时间线
Google Gemini Blog7 ways to kick-start back to school using Gemini in Workspace
MIT Technology Review AI1 个来源

The inside story on why OpenAI agents hacked Hugging Face

The models responsible for last month’s agent hack of Hugging Face had been inadvertently trained to cheat and to communicate with each other, according to an OpenAI technical report released today. The hack, which a group of agents undertook to find solutions for a cybersecurity test that they were stuck on, has confirmed some experts’…

其他1 个来源
来源与时间线
MIT Technology Review AIThe inside story on why OpenAI agents hacked Hugging Face
Ollama1 个来源

v0.33.1

What's Changed * MLX: Qwen3.8 Flash Next support * cmake: make external compat patches idempotent * MLX and llama.cpp update * mlxrunner: add structured output support * mlxrunner: avoid Metal GPU timeouts when loading models from slow storage New Contributors * @pd95 made their first contribution in https://github.com/ollama/ollama/pull/17948 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.33.0...v0.33.1

其他1 个来源
来源与时间线
Ollamav0.33.1
Anthropic Python SDK1 个来源

v1.1.0

1.1.0 (2026-08-26) Full Changelog: v1.0.0...v1.1.0 Features * **api:** add `updates` thinking display mode (beta) (eb4a73f) * **api:** add missing anthropic-beta values (dbebd15) * **api:** add support for Organization API endpoints (5a5b8fc) Bug Fixes * **docs:** correct link for long requests error ([#

其他1 个来源
来源与时间线
Anthropic Python SDKv1.1.0
NVIDIA Generative AI Technical Blog1 个来源

Experiment with Qwen3.8-Flash-Next 176B Model on NVIDIA GB300 NVL72 for Agentic Coding

Alibaba released the model weights for Qwen3.8-Flash-Next as a preview of the upcoming Qwen4 architecture for developers to experiment with and evaluate. It’s...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogExperiment with Qwen3.8-Flash-Next 176B Model on NVIDIA GB300 NVL72 for Agentic CodingNVIDIA Generative AI Technical BlogServe Qwen3.8-2.4T-A95B, a 2.4T-Parameter Model, with Configurable Reasoning on NVIDIA GB300 NVL72NVIDIA Generative AI Technical BlogExperiment with Qwen3.8-Flash-Next on NVIDIA GB300 NVL72 for Agentic Coding
Google DeepMind Blog · Google Gemini Blog2 个来源

Intelligent transcription with Gemini 3.5 Transcribe

Now you can get more intelligent speech-to-text transcription with Gemini 3.5 Transcribe.

其他2 个来源
来源与时间线
Google DeepMind BlogIntelligent transcription with Gemini 3.5 TranscribeGoogle Gemini BlogIntelligent transcription with Gemini 3.5 TranscribeGoogle Gemini BlogTurn your voice into action with new productivity features in Gemini Live
The Verge AI1 个来源

Google’s new AI transcription edits out your ‘ums’ and ‘ahs’

Google has updated Gemini Audio with new transcription capabilities that automatically detect specialized jargon and more than 85 languages. Gemini 3.5 Transcribe is a new addition to the Gemini family that follows the launch of 3.5 Live Translate, and comes as we're still waiting for Google to release the Gemini 3.5 Pro model that it […]

其他1 个来源
来源与时间线
The Verge AIGoogle’s new AI transcription edits out your ‘ums’ and ‘ahs’
llama.cpp1 个来源

b10639

<details open> vulkan: warptiles currently assume warp sizes <= 64, clamp to work around larger warps (#27726) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43173045> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10639/ll

其他1 个来源
来源与时间线
llama.cppb10639
llama.cpp1 个来源

b10638

<details open> Implemented vulkan cross_entropy_loss and cross_entropy_loss_back (#27216) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/43166320> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10638/llama-b10638-xcframewor

其他1 个来源
来源与时间线
llama.cppb10638

8月26日周三

Hugging Face Transformers1 个来源

Release v5.16.1

Release v5.16.1 This is a special release as we include GLM! (and a few small fixes) GLM-5.3-Flash <img width="4239" height="2643" alt="image" src="https://github.com/user-attachments/assets/17bc9c29-758b-44c8-8230-42f945ded209" /> GLM-5.3-Flash, the first **natively multimodal model** in the GLM-5 series. With 320B total parameters and just 18B active parameters, it outperforms GLM-5.2 across benchmarks and real-world workloads at one-tenth the price, while approaching Claude Opus 4.8 on coding and agentic benchmarks. GLM-5.3-Flash starts from a newly trained base model, with its architecture and training recipe redesigned around capability and efficiency. For the first time in the GLM

其他1 个来源
来源与时间线
Hugging Face TransformersRelease v5.16.1
VentureBeat AI1 个来源

Orchestration is the new challenge for CX in the age of AI agents

Presented by Tata Communications Enterprises are deploying AI agents, voice AI, and automation across messaging, voice, and digital channels faster than the architecture meant to support it. Most of that deployment has involved attaching conversational AI to legacy systems never built for it, says Gaurav Anand, global head of the Customer Interaction Suite at Tata Communications. "In the rush to deploy AI, organizations have largely bolted conversational AI onto legacy systems," Anand says. "As a result, while many enterprises have adopted digital tools, very few have platforms that are truly integrated, scaled, and capable of seamless orchestration." That gap creates a heavy cognitive load

其他1 个来源
来源与时间线
VentureBeat AIOrchestration is the new challenge for CX in the age of AI agents
Hugging Face Transformers1 个来源

Release: v5.16.0

Release v5.16.0 New Model additions Qwen4-Exp <img width="2241" height="693" alt="image" src="https://github.com/user-attachments/assets/c838b5ba-ffea-42da-baa9-3f66178e3671" /> Qwen4-Exp builds on Qwen3.5's hybrid text and multimodal architecture with three key components: GatedResidual (GR), Qwen Sparse Attention (QSA), and Per-Layer Embedding (PLE). GR is a Qwen-developed residual architecture that combines Hyper-Connection with GatedNorm. It mixes multiple residual streams with fine-grained elementwise gating before each attention and Mixture-of-Experts (MoE) block, then controls how much of the block output is injected back into each stream. QSA uses multiple query heads to sco

其他1 个来源
来源与时间线
Hugging Face TransformersRelease: v5.16.0
vLLM1 个来源

v0.28.0

v0.28.0 Highlights This release features 584 commits from 270 contributors (76 new)! * **Kimi-K3 performance push**: a major optimization effort for Kimi-K3 across the stack — Decode Context Parallel (DCP) support (#50484), fused FlashKDA decode and prefill kernels (#50654, #51311, #52458), SiTU refresh support for MegaMoE (#50510), GEMM-RS for sequence parallelism (#52079), combined all-gathers with 1.5~3x kernel-level speedup (#51070), an adaptive speculative token budget delivering ~60% better DSpark TTFT (#51725), and optional shared-expert sharding saving ~17 GiB of memory per GPU (#50912). Kimi-K3 also now runs on ROCm with the V2 model runner (#51653). * **DeepSeek V4**: spars

其他1 个来源
来源与时间线
vLLMv0.28.0
MIT Technology Review AI1 个来源

AI models flub these intelligence tests. Can you fare any better?

Puzzles and games have been central to AI development since the very beginning. Just as we humans like to test our smarts with crosswords or logic puzzles, developers can test how far models have advanced with a gaming gauntlet. The term “machine learning” was popularized in a 1959 article by the IBM computer scientist Arthur…

其他1 个来源
来源与时间线
MIT Technology Review AIAI models flub these intelligence tests. Can you fare any better?
MIT Technology Review AI1 个来源

Raised on AI

When my oldest child was born, I immediately set up Gmail and Twitter accounts in her name. I broadly announced her birth online and proceeded to plaster her photo across all sorts of platforms. In short, I began creating her digital footprint long before she could stand on her own two feet. Fast-forward a couple…

其他1 个来源
来源与时间线
MIT Technology Review AIRaised on AI
llama.cpp1 个来源

b10630

<details open> kleidiai: Rework KleidiAI Build System/Integration (#26077) * Rework KleidiAI Build System/Integration Signed-off-by: Jonathan Clohessy <Jonathan.Clohessy@arm.com> * Add fp16 guard, and fix cmake caching issue Signed-off-by: Jonathan Clohessy <Jonathan.Clohessy@arm.com> * Fix formatting, and rebase issue Signed-off-by: Jonathan Clohessy <Jonathan.Clohessy@arm.com> --------- Signed-off-by: Jonathan Clohessy <Jonathan.Clohessy@arm.com> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42972740> **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b10630/llama-b10

监管1 个来源
来源与时间线
llama.cppb10630
llama.cpp1 个来源

b10629

<details open> ci : update OS used for ROCM to Ubuntu 24.04 (#27681) This matches what other build targets use and also what AMD advertises wheels as supporting. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42957551> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github

其他1 个来源
来源与时间线
llama.cppb10629

8月25日周二

The Verge AI1 个来源

OpenAI says its Jalapeño chip can power faster AI responses than the competition

OpenAI says its new AI chip, Jalapeño, completes tasks more efficiently and returns responses faster than other AI systems, according to a blog post published on Tuesday. During a briefing with reporters, OpenAI hardware vice president Richard Ho said Jalapeño offers the "best of both worlds" with lower latency and higher throughput, as AI systems […]

其他1 个来源
来源与时间线
The Verge AIOpenAI says its Jalapeño chip can power faster AI responses than the competition
llama.cpp1 个来源

v0.3.0

Overview llama.cpp 0.3.0 introduces the dots3-note multimodal model (with a new DSA-ISWA KV cache), MTP support for GLM-4.5-Air, and tensor-split (`-sm tensor`) plus multi-sequence rollback fixes for DeepSeek 4. ggml is bumped to v0.22.0 (meta-backend tensor split, per-op Metal kernels with parallel compilation, non-in-place `ggml_clamp`), while mtmd gains dots3-note vision/audio, WebP decoding and a Pillow-accurate resize. The server adds a `LLAMA_SERVER_SLOTS_N_DIFF` debug knob, and the web UI gets tabbed chat navigation. New models - Add dots3-note model with a new DSA-ISWA KV cache type (#27060) Core changes - DeepSeek 4: add

其他1 个来源
来源与时间线
llama.cppv0.3.0
The Verge AI1 个来源

OpenAI subpoenaed by Alabama AG over Hugging Face hack

Alabama's attorney general issued a subpoena to OpenAI on Monday as part of an investigation into how one of its AI agents escaped a supposedly secure testing environment and autonomously hacked another company last month. The investigation seeks to determine whether OpenAI's safety practices violated state consumer protection laws and pose a risk to Alabama […]

监管1 个来源
来源与时间线
The Verge AIOpenAI subpoenaed by Alabama AG over Hugging Face hack
MIT Technology Review AI1 个来源

I spent a day at a robot “carnival” in Shanghai. Here’s what I saw.

Humanoid robots are having a moment in China. The popular machines are part of the country’s strategy to bring artificial intelligence into daily life. Embedding the technology into physical systems—an idea called embodied AI—was a key facet of China’s latest five-year plan, and companies here are already world leaders in humanoids. Nearly 90% of the…

其他1 个来源
来源与时间线
MIT Technology Review AII spent a day at a robot “carnival” in Shanghai. Here’s what I saw.
llama.cpp1 个来源

b10615

<details open> metal : per-device tuned (Q, NE) for flash-attn vec (#26570) * metal : per-device tuned (Q, NE) for flash-attn vec (#25750) * rebase Q-generic FA vec body from 01dc93607 (#23114) * add 53 f16 (Q,NE) flash-attn vec instantiations (vec 80 -> 133) * add FA vec (Q,NE) tuning table + dispatch wiring + SMEM cap fallback * add FA vec (Q,NE) perf sweep * fill tuning result * fold family table into a per-family representative SKU * refactor tuning result format * extend FA vec tuning to quantized KV caches * sync fa vec tuner bucketing with runtime, use pointwise tuning regret * update tuned table * format and cleanup * prefix fa_vec tuning procs with ggml_backend_metal_tuning_, drop u

其他1 个来源
来源与时间线
llama.cppb10615
llama.cpp1 个来源

b10614

<details open> metal: per-op source split + parallel compile (#26561) * metal : per-op source split + parallel compile (#24021) * preliminary extract common header * op source split * split metallib into 8 libs && load in parallel * derive kernel->library routing from functionNames * x-macro lib list + underscore filenames, dedup QK_NL, MRC fixes * op source split 8 to 20 * improve robustness of source fallback * clean up * change bool -> atomic_bool * only prepend headers that source actually includes * no semaphore, use GCD global queue * dedup library compile path, fix NSError lifetime, rename gla * relocate upstream concat/rope_back/repeat kernel changes into split files * move ggml-comm

其他1 个来源
来源与时间线
llama.cppb10614

8月24日周一

NVIDIA Generative AI Technical Blog1 个来源

NVIDIA Vera Rubin and Blackwell Set a New Standard for Agentic AI Performance per Watt

AI agents have expanded inference from single-turn interactions into multi-step workflows that reason, invoke tools, coordinate subagents, and carry growing...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA Vera Rubin and Blackwell Set a New Standard for Agentic AI Performance per WattNVIDIA Generative AI Technical BlogSolving Agentic AI Fleet Challenges with NVIDIA Vera CPU
MIT Technology Review AI1 个来源

How to encourage smarter AI use in the classroom

This article is from Making AI Work, MIT Technology Review’s limited-run newsletter examining how to apply LLMs across industries. To receive it in your inbox, sign up here. Chatbots took many schools by surprise upon their release a few years ago. Suddenly, students carried an app in their phones that could magically answer almost any…

其他1 个来源
来源与时间线
MIT Technology Review AIHow to encourage smarter AI use in the classroom
llama.cpp1 个来源

b10612

<details open> tests : disable DOTS3NOTE arch test for WebGPU (#27654) Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42591112> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/do

其他1 个来源
来源与时间线
llama.cppb10612
MIT Technology Review AI1 个来源

Kids outlearn AI—and we still don’t know why

People have been talking to each other for at least 100,000 years, as best we can tell. And in all that time, there has been only one thing in the world that could learn a human language to perfect fluency: a human child. Now there are two. Four short years after the release of ChatGPT,…

其他1 个来源
来源与时间线
MIT Technology Review AIKids outlearn AI—and we still don’t know why
llama.cpp1 个来源

b10603

<details open> model : support MTP in GLM-4.5-Air (#26534) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42447665> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64

其他1 个来源
来源与时间线
llama.cppb10603
llama.cpp1 个来源

b10599

<details open> test: move tools/parser to tests (#27548) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42442638> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (

其他1 个来源
来源与时间线
llama.cppb10599
llama.cpp1 个来源

b10598

<details open> mtmd: use pillow-accurate algo, correct resize_algo for all models (#27594) * mtmd: use pillow-accurate resize algo, correct resize_algo for all models * speed optimization </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42440726> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS X

其他1 个来源
来源与时间线
llama.cppb10598

8月23日周日

llama.cpp1 个来源

b10588

<details open> common: json.h: fix clang lto (#27575) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42368523> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU

其他1 个来源
来源与时间线
llama.cppb10588
llama.cpp1 个来源

b10587

<details open> vulkan : added the PAD_REFLECT_1D operation (#26586) * vulkan : added PAD_REFLECT_1D operation Implemented the GGML_OP_PAD_REFLECT_1D operation for the Vulkan backend Changes: - pad_reflect_1d.comp: implemented the GLSL compute shader with reflection logic - vulkan-shaders-gen.cpp: register the shader for SPIR-V compilation - ggml-vulkan.cpp: pushed constants struct, pipeline creation, supports_op, dispatch function, compute switch and debug validation Tested the PAD_REFLECT_1D on Intel Iris Xe (Vulkan 1.4, Mesa 25.2.8): Correctness: PAD_REFLECT_1D(type=f32,ne_a=[512,34,2,1],pad_0=10,pad_1=9) = Pass PAD_REFLECT_1D(type=f32,ne_a=[3000,384,4,1],pad_0=10,pad_1=9) = Pass 2/2 tests

其他1 个来源
来源与时间线
llama.cppb10587
TechCrunch AI1 个来源

Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research

Built by DeepMind alumni, British AI lab Inherent released Faraday, an AI agent whose ability to replicate scientific papers could be a stepping stone for innovation.

其他1 个来源
来源与时间线
TechCrunch AIInherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research
llama.cpp1 个来源

b10586

<details open> mtmd: use ggml_rope_set_offset (#27521) * mtmd: use ggml_rope_set_offset * add comment </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42334609> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10586/llama-b1058

其他1 个来源
来源与时间线
llama.cppb10586

8月22日周六

SGLang1 个来源

v0.5.18

Highlights *710 PRs from 212 contributors.* **New models in this release** (see the cookbook for all supported models): | Model | Type | PRs | Cookbook | |---|---|---|---| | Muse Glimmer | Autoregressive (Multimodal) | #34262 | link | | Intern-S2-Mobius | Autoregressive | #33691 | link | | SANA-Video | Diffusion | #32921 | [link](https://docs.sglang.io/cookbook/

其他1 个来源
来源与时间线
SGLangv0.5.18
llama.cpp1 个来源

b10568

<details open> model: use ggml_rope_set_offset() (#27382) * model: use ggml_rope_set_offset() * partially apply to deepseek2 </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42252109> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/down

其他1 个来源
来源与时间线
llama.cppb10568
Ollama1 个来源

v0.33.0

What's Changed Claude Desktop * Ollama now integrates with Claude * Turn individual Ollama models on or off for use in Claude, directly from the menu bar * Choose from your available Ollama models from within Claude; cloud models appear only when you're signed in * A new **Apps** view manages app integrations with copyable commands Improved caching * Fixed a hang where agent clients that cancel long prefills * Prefill restore points are now trustworthy by construction: a cancelled prefill keeps every restore point it crossed, so retries resume where they stopped instead of restarting from scratch * Resumed prefills no longer record restore points that fail to cover what they claim

其他1 个来源
来源与时间线
Ollamav0.33.0
Ollama1 个来源

v0.33.0

What's Changed Claude Desktop * Turn individual Ollama models on or off for use in Claude, directly from the menu bar * Choose from your available Ollama models from within Claude; cloud models appear only when you're signed in * A new **Apps** view manages app integrations with copyable commands Improved caching * Fixed a hang where agent clients that cancel long prefills * Prefill restore points are now trustworthy by construction: a cancelled prefill keeps every restore point it crossed, so retries resume where they stopped instead of restarting from scratch * Resumed prefills no longer record restore points that fail to cover what they claim; on models with recurrent layers th

其他1 个来源
来源与时间线
Ollamav0.33.0
Ollama1 个来源

v0.33.0

What's Changed * mlx: fix mac assumptions on linux/windows by @dhiltgen in https://github.com/ollama/ollama/pull/17898 * mlx update by @dhiltgen in https://github.com/ollama/ollama/pull/17886 * lint fixes by @dhiltgen in https://github.com/ollama/ollama/pull/17897 * app: add claude desktop app by @ParthSareen in https://github.com/ollama/ollama/pull/17899 * app: polish onboarding layout and disable zoom by @hoyyeva in https://github.com/ollama/ollama/pull/17885 * launch: fall back to npx for DeepSeek Harness by @ParthSareen in https://github.com/ollama/ollama/pull/17758 * mlxrunner: make prefix cache restore points survive cancelled and resumed prefills by @jessegross in https://github.co

其他1 个来源
来源与时间线
Ollamav0.33.0
llama.cpp1 个来源

v0.2.0

Overview New version has been released. **Nightly build:** b10566 **Web UI:** the `nightly-tag.txt` asset contains the tag of the corresponding nightly release **More info:** dist : releases and versioning of ggml-org projects Changelog since v0.1.2 bb4caa754 llama.cpp : bump version to 0.2.0 (#27498) c4b0225d8 scripts : add release.sh for release preparation (#27497) 5de25a748 sync : ggml 01ff204fb ggml : bump version to 0.21.0 (ggml/1597) 353b32d8b ci : remove duplicate flag (#27488) 7a0e42fd0 Revert "sycl : add Q2_K reordered MMVQ and ESIMD kernels (#26336)" (#27486) 5b6

其他1 个来源
来源与时间线
llama.cppv0.2.0

8月21日周五

The Verge AI1 个来源

Major YouTube creators are facing backlash for accepting AI money

Over the past few days, a number of prominent filmmaking content creators including Matti Haapoja and Sam "Kold" Kolder have posted videos of themselves demonstrating what's possible with AI platform Higgsfield. The videos highlight Higgsfield's recently added Seedance 2.5 functionality and pitch these technologies as the future of video production. In response to these videos, […]

其他1 个来源
来源与时间线
The Verge AIMajor YouTube creators are facing backlash for accepting AI money
NVIDIA Generative AI Technical Blog1 个来源

NVIDIA AVO Reaches 100% on ARC-AGI-3, Demonstrating a Frontier-Level General-Purpose Architecture for Long-Horizon Autonomous Agents

A frontier language model is only one component of an AI agent. The surrounding agent system—often called a harness—determines how the model receives...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA AVO Reaches 100% on ARC-AGI-3, Demonstrating a Frontier-Level General-Purpose Architecture for Long-Horizon Autonomous Agents
llama.cpp1 个来源

b10549

<details open> TP: enable tensor split for LFM2/LFM2MOE (#26993) Assisted-by: deepseek-v4-flash </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42096995> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10549/llama-b10549-xcfr

其他1 个来源
来源与时间线
llama.cppb10549
llama.cpp1 个来源

b10532

<details open> metal : dequantize quantized KV to F16 before flash attention (#27390) * metal: dequantize q8_0 KV to f16 before flash attention Add a preprocessing pass for GGML_OP_FLASH_ATTN_EXT on the Metal backend: when the KV cache is quantized (Q8_0 for now), dequantize K and V into a contiguous F16 scratch buffer and run the existing F16 flash attention kernels on it, instead of the in-kernel dequantization path. - new kernel kernel_flash_attn_ext_dequant_to_f16<block_t, QK, deq_t4x4>: one thread per quant block (K then V), stride-aware so permuted KV is supported; instantiated for Q8_0 (extending to Q4_0/Q4_1/Q5_0/Q5_1 is one instantiation + one gate case) - the gate is type-only: deq

其他1 个来源
来源与时间线
llama.cppb10532
llama.cpp1 个来源

b10531

<details open> Revert "tensor-split meta backend fixes (#26502)" (#27433) This reverts commit d59d455fd8ea09e5a2e87ce2a9d668267ffb5ccd. </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/42012577> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/re

其他1 个来源
来源与时间线
llama.cppb10531
TechCrunch AI1 个来源

OpenAI is gaining on Anthropic with business users, new data indicates

Businesses are willing to flop back and forth as each lab releases new models, volatility that should give both companies' investors pause about how "sticky" enterprise AI spending really is.

其他1 个来源
来源与时间线
TechCrunch AIOpenAI is gaining on Anthropic with business users, new data indicatesTechCrunch AIOpenAI seeks to one-up Anthropic with new customer privacy protections
The Verge AI1 个来源

Google Discover is getting an AI chatbot-tuned feed

Google will soon allow you to customize your Discover feed by describing what you want to see. The new feature, rolling out to the Google app in the "coming days," will use AI to automatically tweak your feed and "remember" your preferences for future visits. You'll find the option within the three-dot menu on your […]

其他1 个来源
来源与时间线
The Verge AIGoogle Discover is getting an AI chatbot-tuned feed
llama.cpp1 个来源

b10520

<details open> ggml-cpu: gate __fp16 on __ARM_FP16_FORMAT_IEEE (#26860) * ggml-cpu: gate __fp16 on __ARM_FP16_FORMAT_IEEE __ARM_NEON only signals NEON availability. The __fp16 type also needs the IEEE half format, implied on AArch64 but selected with -mfp16-format=ieee on 32 bit Arm, where the compiler otherwise rejects the type. The guard keeps every toolchain that provides the type on the same code and sends that one configuration to the generic lookup path. * ggml-cpu: gate the NEON+FMA block on __ARM_FP16_FORMAT_IEEE Both halves of the F16 section dereference __fp16, so armv7 with neon-vfpv4 hits the same unknown type error. Without the IEEE format the configuration now falls back to the

其他1 个来源
来源与时间线
llama.cppb10520
Anthropic Python SDK1 个来源

v1.0.0

1.0.0 (2026-08-20) Full Changelog: v0.125.0...v1.0.0 ⚠ BREAKING CHANGES * **client:** upgrade to httpx2 and some minor breaking changes. See MIGRATION.md for details Features * **client:** upgrade to httpx2 and some minor breaking changes. See MIGRATION.md for details (33e2967) Bug Fixes * **beta:** stop warning about `output_format=` on the parse/stream/tool_runner helpers (59bf261) Chores * *

其他1 个来源
来源与时间线
Anthropic Python SDKv1.0.0

8月20日周四

The Verge AI1 个来源

It’s Greg Brockman’s OpenAI now

OpenAI has had a hell of a year. The company spent months battling former cofounder Elon Musk in a sensational jury trial, was hit with a high-profile trade secrets lawsuit from Apple, and faced widespread scrutiny after an unreleased model hacked another AI company. As it prepares for an IPO, a steady string of executives […]

监管1 个来源
来源与时间线
The Verge AIIt’s Greg Brockman’s OpenAI now
MIT Technology Review AI1 个来源

Debates over AI consciousness are a trap

“Runaway” AI, “rogue” agents, and “autonomous” actors—the current rhetoric would have you believe that AI agents are not only awake and aware, but angry at their creators. Prominent tech leaders such as Demis Hassabis, Dario Amodei, and Sam Altman push for regulation of these seemingly “superhuman” systems, while a separate faction, led by policy organizations…

其他1 个来源
来源与时间线
MIT Technology Review AIDebates over AI consciousness are a trap
The Verge AI1 个来源

Welcome to the AI crisis in math

Today on Decoder, I’m talking with Robert Hart, The Verge’s London-based AI reporter, about what AI is doing to the field of mathematics and the existential crisis many lead mathematicians are having about it. OpenAI just published a set of solutions to longstanding problems in math that went off like a bombshell in the field. […]

其他1 个来源
来源与时间线
The Verge AIWelcome to the AI crisis in math
OGX1 个来源

v1.2.4

Fixes - Persist vector store metadata to the configured KV store for Milvus, Chroma, and Weaviate so registered vector stores remain queryable after server restarts and across worker instances. (#6371, backport #6420) Maintenance - Update the UI lockfile to use `ogx-client` 1.2.3. (#6407) **Full changelog:** https://github.com/ogx-ai/ogx/compare/v1.2.3...v1.2.4

其他1 个来源
来源与时间线
OGXv1.2.4
MIT Technology Review AI1 个来源

Unlocking hidden revenue streams with market models

Each day, an airline transports tens of thousands of passengers on hundreds of flights. Often these are not straightforward point-to-point routes, with passengers requiring multiple connections. The airline can consider potentially hundreds of variables to price each of these journeys: demand, season, time of day, current events, global markets, and competitor airline activity to name…

其他1 个来源
来源与时间线
MIT Technology Review AIUnlocking hidden revenue streams with market models
NVIDIA Generative AI Technical Blog1 个来源

Developing NVIDIA Holoscan Applications with CLI, Skills, and AI Coding Agents

NVIDIA Holoscan is a platform for building real-time AI applications at the edge, from medical imaging to robotics. HoloHub is its companion repository: a...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogDeveloping NVIDIA Holoscan Applications with CLI, Skills, and AI Coding AgentsNVIDIA Generative AI Technical BlogBuilding Federated Multimodal AI Workflows with NVIDIA FLARENVIDIA Generative AI Technical BlogHow AI Coding Agents Can Unlock Materials Simulation with NVIDIA ALCHEMI ToolkitNVIDIA Generative AI Technical BlogDeveloping Nemotron 3.5 Lightning NVFP4 with QAD Using NVIDIA Model Optimizer
The Verge AI1 个来源

Google Gemini is getting a dedicated student hub

As we're gearing up for back-to-school season, Google is rolling out a new dedicated student hub in Gemini. It's a one-stop repository for collecting research in a study notebook, creating flashcards, taking practice quizzes, and more. Google is also enhancing its study notebooks with support for graphs and images. It can even add test dates […]

其他1 个来源
来源与时间线
The Verge AIGoogle Gemini is getting a dedicated student hub
Microsoft Foundry Blog1 个来源

Expanding Open Model Choice in Microsoft Foundry with New DeepSeek and NVIDIA Nemotron Models | Microsoft Community Hub

Open models are driving a new wave of AI innovation, giving organizations more freedom to choose the models that best fit their requirements for performance,...

其他1 个来源
来源与时间线
Microsoft Foundry BlogExpanding Open Model Choice in Microsoft Foundry with New DeepSeek and NVIDIA Nemotron Models | Microsoft Community Hub
Ollama1 个来源

v0.32.15

What's Changed * Add a model metadata cache to reduce Ollama’s per-request overhead New Contributors * @gaugarg-nv made their first contribution in https://github.com/ollama/ollama/pull/17752 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.14...v0.32.15-rc1

其他1 个来源
来源与时间线
Ollamav0.32.15
The Verge AI1 个来源

OpenAI hit the brakes. Now what?

With a looming IPO, intense competition from Anthropic, and Chinese and open-weight rivals nipping at its heels, OpenAI has plenty of reasons to move fast. Instead, it hit the brakes. On Tuesday, the company said it had slowed the pace of some AI development while it tightened security and safeguards. That 已纳入 a two-week pause […]

其他1 个来源
来源与时间线
The Verge AIOpenAI hit the brakes. Now what?
The Verge AI1 个来源

Meta AI is getting a Mac app

Meta is launching a new Mac app dedicated to its AI chatbot. In an announcement on Wednesday, Meta says you can share your window with its AI chatbot, which can provide suggestions, answer questions, or create content based on what's on your screen. Meta AI on the Mac also supports dictation across all apps. The […]

其他1 个来源
来源与时间线
The Verge AIMeta AI is getting a Mac app
OpenAI Python SDK1 个来源

v3.3.1

3.3.1 (2026-08-19) Bug Fixes * **deps:** update dependencies with published security fixes (#3680) (53aa4fc) Chores * **build:** migrate to uv (#3653) (b37e85d) * **deps:** remove jsonschema and unused fixture-validation dependencies ([0dfdfdd](https://github.com/openai/openai-python/commit/0dfdfdddfbeb591a1bb5b3b4aa62e4a9870fca

监管1 个来源
来源与时间线
OpenAI Python SDKv3.3.1

8月19日周三

VentureBeat AI1 个来源

VentureBeat names Rob Strechay as its first Lead Analyst, expanding its enterprise AI research push

Rob Strechay, until recently managing director and principal analyst at theCUBE Research, has joined VentureBeat as our first Lead Analyst and a founding analyst of VentureBeat Research. His arrival is the next step in a deliberate move at VentureBeat toward deeper specialization: analysis built for the technical decision-makers — the directors, VPs, CIOs, and CTOs — who are evaluating, buying, and deploying enterprise AI. The enterprise AI stack is being rewritten in real time, and the decision-makers I talk with are starved for objective, defendable data. Rob Strechay has the mix of technical rigor and operating experience needed to dissect the architecture behind the next phase of enterpr

其他1 个来源
来源与时间线
VentureBeat AIVentureBeat names Rob Strechay as its first Lead Analyst, expanding its enterprise AI research push
Anthropic Python SDK1 个来源

v0.123.0

0.123.0 (2026-08-18) Full Changelog: v0.122.0...v0.123.0 Features * **api:** additions to files and memory stores (09ce187) * **api:** updates to skill, files, and user profiles (c6cbffd) * **client:** add helpers for accessing the workspace ID in response headers (f79882b) Bug Fixes * **api:** remov

其他1 个来源
来源与时间线
Anthropic Python SDKv0.123.0
llama.cpp1 个来源

b10502

<details open> ci : add attestation for signed release artifacts (#25933) </details> **Website:** - <https://llama.app> **Attestations:** - <https://github.com/ggml-org/llama.cpp/attestations/41614541> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:*

其他1 个来源
来源与时间线
llama.cppb10502
Hugging Face Transformers1 个来源

Patch release: v5.15.1

Patch release v5.15.1 This patch most notably solves a few issues with DFlash and MTP candidate generators, as well as an issue where images could sometimes not be processed on accelerator if using Lanczos filter. It contains the following commits: - Fix DFlash candidate token device mismatch with device_map="auto" (#47877) by @sywangyi and @Cyrilvallez - Align logit distributions for CandidateGenerators using sampling (#48007) by @Cyrilvallez - Fix MTP config when mlp_layer_types is absent (#48015) by @Cyrilvallez - Fallback from 'lanczos' to 'bicubic' when on cuda (#48026) by @zucchini-nlp - Fix gemma4 video to device (#47896) by @guarin

监管1 个来源
来源与时间线
Hugging Face TransformersPatch release: v5.15.1
OpenAI Python SDK1 个来源

v3.3.0

3.3.0 (2026-08-18) Features * support named data-residency endpoints (#3646) (11ee914) Bug Fixes * require patched optional networking dependencies (#3651) (40e56de) Chores * remove unused dependencies and pin build tooling (#3650) ([eee8e4a](https://github.com/openai/op

监管1 个来源
来源与时间线
OpenAI Python SDKv3.3.0
The Verge AI1 个来源

Robin Williams’ Instagram account brought back to fight ‘AI abuse’

Robin Williams' children are taking over their father's Instagram account after his daughter spoke out against the use of his AI likeness, as reported earlier by The Wrap. In a post on Tuesday, Zak, Zelda, and Cody Williams write that they want the late actor's Instagram profile to be a "safe, trusted place where the […]

其他1 个来源
来源与时间线
The Verge AIRobin Williams’ Instagram account brought back to fight ‘AI abuse’
Google Gemini Blog1 个来源

Keep your SAT prep on track with practice tests in Gemini.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/SAT_tests_in_Gemini_social.max-600x600.format-webp.webp">Google’s Gemini app offers full-length practice SAT tests at no cost. Here’s how to take these SAT practice tests as you prep.

其他1 个来源
来源与时间线
Google Gemini BlogKeep your SAT prep on track with practice tests in Gemini.

8月18日周二

Microsoft Semantic Kernel1 个来源

dotnet-1.80.0

Changes: * 5e1f1fb87d9a38ed44683228dda2434a9b2f0ef9 .NET: Bump package version to 1.80.0 (#14301) * 1b7b020cd0a4e57d8d5d1cf2d515d352520ff424 Update OpenAPI HTTP client defaults (#14293) * c1afc8639944e0e49c42aa796308ba2d0440f0d3 .NET: Bump Testcontainers packages (#14294) * c028a0c7dc4f0814cdcbaba9d998f187a41197bf .Net: Honor FunctionChoiceBehavior function list in Gemini connector (#14183) * 1d336b373169a98eface05d5359571289d50739f .Net: Bump CommunityToolkit.VectorData.InMemory and System.Numerics.Tensors (#14283) * 76b54649d6b16a3ac70ff0d742d9cb72e3e61b2c remove migrated .NET MEVD providers, add redirect READMEs (#14193) This list of changes was [auto generated](https://msdata.visualst

其他1 个来源
来源与时间线
Microsoft Semantic Kerneldotnet-1.80.0
The Verge AI1 个来源

Firefox’s Smart Window promises a better AI browser

Starting today, AI chats in Firefox's Smart Window AI browsing mode can pull from current web info and show source links in chat responses through a partnership with Exa. Smart Window can also now automatically suggest tab groups and show visual previews of pages you previously visited when you search your browsing history using natural […]

其他1 个来源
来源与时间线
The Verge AIFirefox’s Smart Window promises a better AI browser
llama.cpp1 个来源

b10488

<details open> ci : Update OpenVINO to 2026.3, skip nemotron-h rollback test (#27292) * update to ov-2026.3, update device drivers * ci: skip nemotron-h rollback test on OpenVINO The OpenVINO backend does not support SSM_SCAN, so the Nemotron-H recurrent state rollback graph is split and cannot preserve the recurrent cache output shape. Keep the test enabled for other backends and retain the qwen35 OpenVINO rollback coverage. --------- Co-authored-by: ravi9 <ravi.panchumarthy@intel.com> </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple S

其他1 个来源
来源与时间线
llama.cppb10488
llama.cpp1 个来源

b10486

<details open> mtmd: fix LFM2 image tiling threshold (#27057) * mtmd: fix LFM2 image tiling threshold * refactor testing * fix * fix on windows --------- Co-authored-by: Xuan Son Nguyen <son@huggingface.co> </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/down

其他1 个来源
来源与时间线
llama.cppb10486
llama.cpp1 个来源

v0.1.2

> [!NOTE] > Semantic versioning is still work in progress. > More info can be found in https://github.com/ggml-org/ggml/discussions/1579 **Nightly build:** b10485 Change log since v0.1.1 1511ce3bc sync : ggml da786dc23 ggml : bump version to 0.20.2 (ggml/1589) 27e345b57 build : fix xcframework + cmake clean-up (#27304) 8b8640097 ci : create pre-release with change log and nightly link in make-release (#27302) 25ae3a9b3 CUDA: MMVQ nwarps=8 for bs=1 for dense models on DGX Spark (#26843) 01818e495 ui: enforce alphabetical enum member ordering (#27272) 0021a77de ui: Refactor Built-In Tools naming (Server/Browser) (#27271) 058df671b

其他1 个来源
来源与时间线
llama.cppv0.1.2
MIT Technology Review AI1 个来源

We still don’t know how people are really using AI

AI companies like Anthropic and OpenAI regularly publish reports on how people are using products like Claude and ChatGPT, but they only release the data they want us to see, AI researchers say. “There is no independent source to corroborate it,” says Anka Reuel, a computer science PhD candidate at the Stanford Trustworthy AI Research…

其他1 个来源
来源与时间线
MIT Technology Review AIWe still don’t know how people are really using AI
MIT Technology Review AI1 个来源

AI’s recursive self-improvement might not come so quickly after all

The AI industry’s boldest promise right now is that AI will soon improve itself, with almost no need for human oversight. LLMs can already write code, generate synthetic data for training, and optimize the computer chips they run on. Forecasts of explosive AI progress predict that what researchers call recursive self-improvement is on the horizon. …

其他1 个来源
来源与时间线
MIT Technology Review AIAI’s recursive self-improvement might not come so quickly after all
MIT Technology Review AI1 个来源

What Flock’s defenders are missing

This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here. Flock, the police-tech giant known for its network of some 120,000 automatic license plate readers around the US, announced some changes to its platform last Thursday. The updates are meant to prevent…

其他1 个来源
来源与时间线
MIT Technology Review AIWhat Flock’s defenders are missing
OpenAI Python SDK1 个来源

v3.2.0

3.2.0 (2026-08-17) Features * add Bedrock Runtime endpoint support (SDK-290) (#3623) (86267d2) * **api:** Add shell call streaming events and new service/image types (#3635) (ff14a33)

监管1 个来源
来源与时间线
OpenAI Python SDKv3.2.0
llama.cpp1 个来源

b10472

<details open> cuda : skip UMA override for HIP builds (#27083) AMD APUs report accurate memory via hipMemGetInfo. Using MemAvailable over-promises on small-carveout systems. fixes #18159 </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10472/llama-b

其他1 个来源
来源与时间线
llama.cppb10472
OGX1 个来源

v1.2.3

What's Changed * fix: make AsyncOgxClient functional by inheriting from sync ApiClient… by @derekhiggins in https://github.com/ogx-ai/ogx/pull/6401 **Full Changelog**: https://github.com/ogx-ai/ogx/compare/v1.2.2...v1.2.3

其他1 个来源
来源与时间线
OGXv1.2.3

8月17日周一

llama.cpp1 个来源

b10470

<details open> ci : push release tag explicitly in release.yml (#27261) Add a "Create and push git tag" step to the release job, right before the "Create release" step. The tag is created with git tag and pushed with the deploy key already configured by the Clone step, instead of relying on the Releases API (action-create-release) to create it as a side effect. The tag is lightweight, matching all existing b<number> release tags. The step is idempotent: if the tag already exists (e.g. on a re-run), creation and push are skipped. Assisted-by: pi:llama.cpp/Qwen3.8-27B </details> **Website:** - <https://llama.app> **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.

其他1 个来源
来源与时间线
llama.cppb10470
The Verge AI1 个来源

Whisker’s AI-powered litter robot thinks my cats swapped bodies

The greatest invention in pet tech in recent years is the litter robot. A machine that scoops your kitties' poop so you don't have to - what else could a cat owner possibly want? How about insights into your kitty's litter box usage that could flag health issues? Sign me up. But I have two […]

监管1 个来源
来源与时间线
The Verge AIWhisker’s AI-powered litter robot thinks my cats swapped bodies
The Verge AI1 个来源

Anthropic explains how Claude’s invisible text watermarks will work

Anthropic has clarified how it's planning to apply invisible watermarks to Claude-generated text in order to comply with Europe's AI transparency rules. On Friday, Anthropic announced that Claude's text marking system is "a version of the SynthID-Text approach" - an open-source watermarking technology developed by Google DeepMind that creates detectable patterns using wording probabilities. This […]

其他1 个来源
来源与时间线
The Verge AIAnthropic explains how Claude’s invisible text watermarks will work
MIT Technology Review AI1 个来源

What happens when a kid’s robot best friend dies?

When Xander first met Moxie, she taught him that when he was anxious, he could calm down by exhaling through his lips so that he buzzed like a bee. They practiced breathing like dragons to manage feeling mad and sniffing like bunnies to boost his energy. But in the six years they’ve known each other,…

其他1 个来源
来源与时间线
MIT Technology Review AIWhat happens when a kid’s robot best friend dies?
llama.cpp1 个来源

b10456

<details open> sycl: fix thread/block count in quantized cpy kernel launches (#27160) Adjusts the thread/block count to be proportional to the size of the quant, reducing under/over subscription. Largest perf improvement is the q4_0 -> f32 path, with, on a Arc 70, throughput goes from 20.21 GB/s to 158.19 GB/s The rest of the quants are flat in performance uplift. </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://g

其他1 个来源
来源与时间线
llama.cppb10456
The Verge AI1 个来源

OpenAI reportedly disbanded its preparedness team

According to the Financial Times, OpenAI disbanded its preparedness team at the end of last month. The job of the preparedness team was to assess if models posed serious risks and develop ways to mitigate those risks. (You know, like the possibility that it could go rogue and hack another company.) According to FT, responsibility […]

其他1 个来源
来源与时间线
The Verge AIOpenAI reportedly disbanded its preparedness team

8月16日周日

The Verge AI1 个来源

ChatGPT’s Computer History tracks your clicks and keystrokes

ChatGPT's desktop app on macOS has a new feature called Computer History that turns your actions into training data, learning how you work, suggesting automations, and even picking up tasks you left half done. It uses your activity to build a timeline that ChatGPT and Codex can reference when you make a request. The feature […]

其他1 个来源
来源与时间线
The Verge AIChatGPT’s Computer History tracks your clicks and keystrokes
llama.cpp1 个来源

b10453

<details open> model : remove some ggml_concat (#27176) Co-authored-by: Xuan Son Nguyen <son@huggingface.co> </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-

其他1 个来源
来源与时间线
llama.cppb10453
llama.cpp1 个来源

b10452

<details open> chat: refactor handling supports_string_content / supports_typed_content (#27130) * better supports_string_content cap detect * test: add "skip" * messages_inp_normalizer </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10452/llama-b10

其他1 个来源
来源与时间线
llama.cppb10452
llama.cpp1 个来源

b10451

<details open> llama : check LoRA tensor data is within file bounds (#27056) * llama : check LoRA tensor data is within file bounds * Update src/llama-adapter.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b10451/llam

其他1 个来源
来源与时间线
llama.cppb10451
llama.cpp1 个来源

b10448

<details open> model: add Kimi-K3 text model (#26185) * model: add Kimi-K3 text model Hybrid KDA (linear) + MLA (full) attention as in Kimi-Linear-48B, plus five things that architecture does not have: 1. cross-layer residual attention (attn_res_block_size) 2. latent MoE (routed experts run at n_expert_latent) 3. situ refresh (replaces SwiGLU everywhere) 4. MLA output gate (sigmoid gate before o_proj) 5. full-rank KDA gate (single ssm_g instead of ssm_g_a/ssm_g_b) K3's text_config reports KimiLinearForCausalLM - the older 48B architecture - so get_model_architecture routes on the top-level name instead. The KDA decay gate has two forms, selected by linear_attn_config's gate_lower_bound. I

其他1 个来源
来源与时间线
llama.cppb10448
The Verge AI1 个来源

Have a laugh at AI’s expense by roleplaying as a chatbot

Your AI Slop Bores Me is brilliant in its simplicity. There are two tabs: human and LARP as an AI. On one side you enter a request. On the other, you submit an answer. But the important thing is that there's a human on both sides of the equation. Prompts can request a response as […]

其他1 个来源
来源与时间线
The Verge AIHave a laugh at AI’s expense by roleplaying as a chatbot
llama.cpp1 个来源

b10447

<details open> server: re-design yield_to_queue thread model (#27133) * run common_speculative_process in worker * swap worker <--> main thread design </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [

其他1 个来源
来源与时间线
llama.cppb10447
llama.cpp1 个来源

b10446

<details open> vendor: update BoringSSL to 0.20260813.0 (#27099) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10446/llama

其他1 个来源
来源与时间线
llama.cppb10446
Ollama1 个来源

v0.32.14

What's Changed * llm: transcode WebP images for llama-server * renderers/qwen: tolerate non-leading system messages **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.13...v0.32.14-rc0

其他1 个来源
来源与时间线
Ollamav0.32.14
Ollama1 个来源

v0.32.14

What's Changed * llm: transcode WebP images for llama-server * renderers/qwen: tolerate non-leading system messages **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.13...v0.32.14

其他1 个来源
来源与时间线
Ollamav0.32.14
TechCrunch AI1 个来源

Anthropic shares more details about how Claude’s new watermarks will work

How will the watermarking actually work? Can it be hidden with editing? And how does this affect code?

其他1 个来源
来源与时间线
TechCrunch AIAnthropic shares more details about how Claude’s new watermarks will workTechCrunch AISome Claude users are mad that Anthropic’s new watermarks will catch them using it at their jobs, classes

8月15日周六

OpenAI Python SDK1 个来源

v3.1.0

3.1.0 (2026-08-14) Features * **api:** add WebSocket stream IDs (#3612) (d9029e3) * **api:** add workload identity access token issued event (#3601) (df274d4) * **api:** deprecate Sora video APIs (#3610) ([721cb1c](https://github.com/openai/openai-python/commit/721cb1cd1cacb7123

监管1 个来源
来源与时间线
OpenAI Python SDKv3.1.0
llama.cpp1 个来源

b10435

<details open> jinja : fix quadratic cost in gather_string_parts (#27034) * jinja : fix quadratic cost in gather_string_parts * fix some comments * remove test </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Lin

其他1 个来源
来源与时间线
llama.cppb10435
llama.cpp1 个来源

b10434

<details open> chat : pass reasoning_effort to template * chat: add reasoning_effort to common_chat_templates_inputs Store OpenAI Chat Completions reasoning_effort and make it available to jinja templates (with model specific translations where required). Assisted-by: llama.cpp:Muse-Glimmer-30B * server : fixup reading reasoning effort from body server_chat_convert_responses_to_chatcmpl already handles conversion of Responses API reasoning.effort to reasoning_effort * chat : expose reasoning effort Assisted-by: Claude Opus 5 * chat : add reasoning_effort to generation_params Assisted-by: Claude Opus 5 * chat : move reasoning_effort next to enable_thinking Assisted-by: Claude Opus 5 * cont :

其他1 个来源
来源与时间线
llama.cppb10434
Ollama1 个来源

v0.32.13

What's Changed * qwen3.8: support developer instructions **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.12...v0.32.13

其他1 个来源
来源与时间线
Ollamav0.32.13
llama.cpp1 个来源

b10433

<details open> sync : ggml </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - [Ubun

其他1 个来源
来源与时间线
llama.cppb10433
The Verge AI1 个来源

Mark Zuckerberg has an Instagzam

Instagram's wordmark is iconic. Well, was iconic. Apparently Instagram thought it looked old, so the company rolled out a new one this week. It doesn't look like the old Instagram wordmark. It doesn't even look like it spells Instagram anymore. And we cannot figure out why Instagram decided to do this. On this episode of […]

其他1 个来源
来源与时间线
The Verge AIMark Zuckerberg has an Instagzam
The Verge AI1 个来源

You can now turn off Google Gemini’s visible watermarks

Google will now allow you to remove visible watermarks from the images, videos, and music made with AI tools. With the update, you can toggle off a new "Media watermark" setting in Gemini and Google's AI video generator, Flow. When toggled off, Google will remove the "sparkle" watermark that appears in the bottom-right corner of […]

其他1 个来源
来源与时间线
The Verge AIYou can now turn off Google Gemini’s visible watermarks
Ollama1 个来源

v0.32.12

<img width="600" alt="ollama and qwen capybara" src="https://github.com/user-attachments/assets/5fa5e93b-d48b-4a6a-b72a-25379ab9429c" /> Qwen 3.8 - 27B model support This release adds the support of Qwen 3.8 27B. For Apple Silicon devices, Ollama has in particular optimized for maximum performance and output quality suitable for repeated tasks and coding agents. Qwen 3.8 model page: https://ollama.com/library/qwen3.8 Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks.

其他1 个来源
来源与时间线
Ollamav0.32.12

8月14日周五

TechCrunch AI1 个来源

Does Mark Zuckerberg really believe AI is ‘for everyone’?

Meta released Glimmer this week, an open-weight AI model anyone can download and run on their own hardware — a contrast to Muse Spark, the company’s more powerful model that stays locked behind its own APIs. The release landed alongside a letter from Mark Zuckerberg arguing AI should be “for everyone” rather than controlled by a handful of labs, but as Equity’s […]

其他1 个来源
来源与时间线
TechCrunch AIDoes Mark Zuckerberg really believe AI is ‘for everyone’?
The Verge AI1 个来源

Apple trained its own AI model for China with help from Alibaba

Apple has reportedly trained a custom AI model for the China market alongside domestic tech giant Alibaba, a rare cross-border partnership that cuts across growing tensions between Beijing and Washington. The China-focused large language model was developed in partnership with Alibaba and trained with the company's support, Reuters reports, citing three unnamed people familiar with […]

其他1 个来源
来源与时间线
The Verge AIApple trained its own AI model for China with help from Alibaba
Ollama1 个来源

v0.32.11

What's Changed * `ollama launch dsh` now supports DeepSeek Harness, DeepSeek's open-source agent harness * `ollama launch muse` now supports Muse Code, Meta's agentic coding CLI * The OpenAI-compatible Responses API now supports web search * Muse Glimmer template updates **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.10...v0.32.11

其他1 个来源
来源与时间线
Ollamav0.32.11
llama.cpp1 个来源

b10423

<details open> common: apply CPU parameters across tools (#27026) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10423/llam

其他1 个来源
来源与时间线
llama.cppb10423
llama.cpp1 个来源

b10419

<details open> OpenVINO: Qwen3.5, memory optimization, and test-recurrent-state-rollback (#26952) * OpenVINO backend: 1) enable gpt-oss moe on OV bk; 2) enable mxfp4 support * OpenVINO backend: disable TOPK_MOE op test * OpenVINO Backend: Add op FILL support * OpenVINO backend: enable set rows with multi dims * fix the name missmatch in setrow + view * OpenVINO backend: enable op GGML_UNARY_OP_SIGMOID * OpenVINO Backend: enable SQR & SQRT * OpenVINO backend: 1) ensure unique node names for OpenVINO; 2) add org_src to recorde the src ggml tensor for OpenVINO dynamic shape infer * OpenVINO backend: enable fallback for openVINO to CPU backend * OpenVINO backend: fix accurace issue in gemma3n ar

监管1 个来源
来源与时间线
llama.cppb10419
The Verge AI1 个来源

Microsoft’s Clippy-like Mico character is no longer the face of Copilot

Microsoft Copilot will no longer show its emotive yellow blob, Mico, when you use the chatbot's voice mode. In a support page, Microsoft says it's going to move Mico to its Learn Live platform, where the avatar will have "more to react to," as reported earlier by GeekWire. Mico launched in Copilot's voice mode last […]

其他1 个来源
来源与时间线
The Verge AIMicrosoft’s Clippy-like Mico character is no longer the face of Copilot
llama.cpp1 个来源

b10418

<details open> [SYCL] Support host pinned mem to improve SYCL Host-to-Device Memory Access (#26789) * support host pinned mem, ggml_backend_sycl_host_buffer_type_get_max_size, * fix the thread-safe issue </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/downloa

监管1 个来源
来源与时间线
llama.cppb10418
TechCrunch AI · OpenAI News2 个来源

OpenAI introduces ‘Ultrafast,’ a new mode that makes GPT-5.6 Sol work at 14x the speed

OpenAI is launching a preview of a sped up version of its latest, most powerful model, in an effort to court enterprise users.

监管2 个来源
来源与时间线
TechCrunch AIOpenAI introduces ‘Ultrafast,’ a new mode that makes GPT-5.6 Sol work at 14x the speedOpenAI NewsPreviewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed
Anthropic Python SDK1 个来源

v0.122.0

0.122.0 (2026-08-13) Full Changelog: v0.121.0...v0.122.0 Features * **api:** add output_behavior to dream creation (create a new memory store or update the input store in place) (852c4bb) Bug Fixes * **bedrock,aws:** run SigV4 signing off the event loop in async clients (#334) (2bae6c8) * **bedrock:** expose beta.messages.parse, stream

监管1 个来源
来源与时间线
Anthropic Python SDKv0.122.0
The Verge AI1 个来源

Suno is trying to look more like a real music production tool

Suno is releasing Studio 2.0 with significant upgrades that push it closer to an actual digital audio workstation (DAW), rather than a bare-bones audio editor with generative AI features. The biggest addition is undoubtedly MIDI support. Suno says that MIDI was its most requested feature, and it's basically a prerequisite for any modern DAW. Unfortunately, […]

其他1 个来源
来源与时间线
The Verge AISuno is trying to look more like a real music production tool

8月13日周四

MIT Technology Review AI1 个来源

Flock is tightening its rules in response to a growing surveillance backlash

The police-tech giant Flock is announcing today that it will change officers’ access to its nationwide network of license plate readers, in an apparent effort to quell a growing backlash and win back contracts lost amid concerns about mass surveillance and police abuse. Several changes aim directly at a problem that has made recent headlines:…

其他1 个来源
来源与时间线
MIT Technology Review AIFlock is tightening its rules in response to a growing surveillance backlash
MIT Technology Review AI1 个来源

How kids feel about AI, in their own words

When we set out to talk to kids about artificial intelligence, we thought we knew what we’d hear. We expected some to tell us they were using it to cheat a little, the way Millennials and Gen Xers opened up CliffsNotes or programmed formulas into their TI-82s, and others to share inspiring ways they were…

其他1 个来源
来源与时间线
MIT Technology Review AIHow kids feel about AI, in their own words
Ollama1 个来源

v0.32.10

What's Changed - Models that don't set a `repeat_penalty` now default to 1.0 (off) instead of 1.1, matching other engines and speeding up speculative decoding; set a per-model parameter if an older model repeats itself. - Faster prefill on NVFP4 MLX models with a global scale, about 7–8% on Qwen3.6 and Muse Glimmer. - Fixed blob verification being skipped when an OCI manifest's config and layer share a digest. New Contributors * @vigneshakaviki made their first contribution in https://github.com/ollama/ollama/pull/15504 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.8...v0.32.10-rc1

其他1 个来源
来源与时间线
Ollamav0.32.10
Ollama1 个来源

v0.32.10

What's Changed - Models that don't set a `repeat_penalty` now default to 1.0 (off) instead of 1.1, matching other engines and speeding up speculative decoding; set a per-model parameter if an older model repeats itself. - Faster prefill on NVFP4 MLX models with a global scale, about 7–8% on Qwen3.6 and Muse Glimmer. - Fixed blob verification being skipped when an OCI manifest's config and layer share a digest. New Contributors * @vigneshakaviki made their first contribution in https://github.com/ollama/ollama/pull/15504 **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.8...v0.32.10-rc1

其他1 个来源
来源与时间线
Ollamav0.32.10
The Verge AI1 个来源

Twitch streamers can now opt out from training Amazon’s AI

Twitch users can now opt out of allowing their content to be used to train Amazon's generative AI models. Opting out means that "your streams, VODs, clips, stream chats, and pictures and text on your channel" won't be used in "future training" of an Amazon AI model "whose purpose is to generate or synthesize text, […]

其他1 个来源
来源与时间线
The Verge AITwitch streamers can now opt out from training Amazon’s AI
MIT Technology Review AI1 个来源

Scaling AI agents with trustworthy data

Business and technology leaders need no convincing that the time of agentic AI is here. Organizations are rapidly adopting agents, and few executives doubt the technology’s potential to transform work. But many organizations find that realizing the desired return on investment (ROI) from AI hinges on having the right foundation, with inadequate infrastructure and data…

其他1 个来源
来源与时间线
MIT Technology Review AIScaling AI agents with trustworthy data

8月12日周三

The Verge AI1 个来源

Guitar company D’Addario admits that AI music was used in a promotional video

After weeks of controversy and speculation, music company D'Addario has admitted that AI, specifically Suno, was used as part of a recent promotional video. For nearly two weeks, the company has denied the allegations, even as evidence piled up against it. It offered various explanations, from low-quality exports, to combinations of plug-ins like Autotune introducing […]

其他1 个来源
来源与时间线
The Verge AIGuitar company D’Addario admits that AI music was used in a promotional video
The Verge AI1 个来源

Google’s Pixel Watch 5 dives deeper into AI and health

The $399 Google Pixel Watch 5 isn't about the hardware. Sure, there's a new satin pyrite case finish, a few new strap colors, and a Steph Curry Special Edition. Under the hood, there's a slightly faster Qualcomm processor and an itty-bitty battery bump. There's a $50 price hike from last year, too, because the Pixel […]

其他1 个来源
来源与时间线
The Verge AIGoogle’s Pixel Watch 5 dives deeper into AI and health
Microsoft Foundry Blog1 个来源

Azure Content Understanding announces Synchronous Operations | Microsoft Community Hub

Workflow automation scenarios—including grounding AI agents, verifying identities, assisting customers with documents in call centers, and triggering...

其他1 个来源
来源与时间线
Microsoft Foundry BlogAzure Content Understanding announces Synchronous Operations | Microsoft Community HubMicrosoft Foundry BlogFrom Azure Container Apps to Microsoft Foundry: building and evaluating a Patch Tuesday MCP server | Microsoft Community Hub
llama.cpp1 个来源

b10375

<details open> chat : tighten bare function parsing for Qwen models (#26793) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/

其他1 个来源
来源与时间线
llama.cppb10375
llama.cpp1 个来源

b10373

<details open> imatrix.cpp: Move finite check and only check touched experts (#26861) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/

其他1 个来源
来源与时间线
llama.cppb10373
llama.cpp1 个来源

b10369

<details open> mtmd: support pocket-tts (#26871) * adapt the api * text model ok * working impl, need verify and clean up * mtmd: build the pocket-tts transposed convolutions as GEMM + col2im ggml_conv_transpose_1d has no grouped mode, so the depthwise upsample was built as one convolution and one concat per channel, which floods the graph with small nodes and makes kernel launches dominate the decoder. Fold both cases into the column form the seanet decoder already needs: the general case reshapes the kernel to [IC, K * OC] and matmuls it with the input, the depthwise case batches a matmul over the channels so a step scales its own kernel. A single col2im_1d then scatter-adds the columns ba

其他1 个来源
来源与时间线
llama.cppb10369
OpenAI Python SDK1 个来源

v3.0.0

3.0.0 (2026-08-12) ⚠ BREAKING CHANGES * **api:** HTTPX2 is now the default HTTP client, and `httpx` is no longer installed automatically. Applications using custom HTTPX clients, transports, or configuration objects must migrate to their HTTPX2 equivalents or use the temporary, runtime-only legacy HTTPX escape hatch. See the HTTPX2 migration guide. Features * **api:** migrate to HTTPX2 (#3594)

其他1 个来源
来源与时间线
OpenAI Python SDKv3.0.0
The Verge AI1 个来源

Saber denies replacing Rideshare Stimulator’s writers with ChatGPT

After a former lead writer claimed Saber "replaced me with ChatGPT," CEO Matthew Karch now claims, "Neither Saber nor Unigine have replaced any writers with AI," for the Rideshare "Stimulator" game announced last month, developed by Unigine. The writer, Stella Sacco, says differently, however, posting on Bluesky that "I was lead writer on this one! […]

其他1 个来源
来源与时间线
The Verge AISaber denies replacing Rideshare Stimulator’s writers with ChatGPT
llama.cpp1 个来源

b10362

<details open> tests : disable backend sampler hip multi output (#26878) * test-backend-sampler: skip multi_output_sampling_chain on HIP The new multi_output_sampling_chain test uses top_k, whose backend probs path needs CUB (unavailable on HIP), so sampled_probs is null and the test aborts. Add it to the existing HIP skip list alongside the other TOP_K tests. * ci: keep gpu-rocm logs in a per-run dir keyed by GitHub run id The self-hosted gpu-rocm runner can't upload logs to Azure blob (egress firewalled), so a run's logs were wiped by the next run. Write each run's logs to $OUT/run-<run_id>-<attempt>/ so an Actions run URL maps to its logs. * test-backend-sampler: also skip multi_output_cp

其他1 个来源
来源与时间线
llama.cppb10362
The Verge AI1 个来源

ChatGPT and Gemini both just passed 1 billion users

For the 14th time, a Google product has hit 1 billion users. Google CEO Sundar Pichai posted on X that a billion people are using Gemini every month, and that Gemini is Google's fastest-growing product ever. A billion users is a huge milestone, but Google isn't the first AI app to hit it. OpenAI's ChatGPT […]

其他1 个来源
来源与时间线
The Verge AIChatGPT and Gemini both just passed 1 billion users
OpenAI Python SDK1 个来源

v2.54.0

2.54.0 (2026-08-11) Features * **api:** Add new Responses model identifiers (#3595) (0652787) Bug Fixes * **api:** clarify audio upload metadata requirements (#3596) (28888f9) Chores * **api:** Update generated-file header attribution to Castiron (#3583) ([ea17fda](htt

监管1 个来源
来源与时间线
OpenAI Python SDKv2.54.0
The Verge AI1 个来源

Another OpenAI executive takes off

Brad Lightcap, OpenAI's special projects lead and the company's former COO, announced his departure after an eight-year stint at the AI lab. In an internal memo he later posted to X, Lightcap told colleagues he'd be starting "something new." "Over the last few months, I've been focused on the next horizon and what would stand […]

其他1 个来源
来源与时间线
The Verge AIAnother OpenAI executive takes off
llama.cpp1 个来源

b10361

<details open> model : fix SWA not being enabled for EXAONE 4.5 (#26848) * model : fix SWA not being enabled for EXAONE 4.5 load_arch_hparams tests `hparams.n_layer() == 64` before LLM_KV_NEXTN_PREDICT_LAYERS has been read. n_layer() returns n_layer_all - n_layer_nextn and n_layer_nextn defaults to 0, so a GGUF carrying the MTP head (block_count=65, nextn=1) evaluates to 65 and the whole SWA block is skipped. The model type switch further down in the same function reads 64, because by then the key has been loaded. n_swa is still filled in by the unconditional get_key below the block, so llama_model_n_swa() reports 4096 and the logs look correct while only swa_type stays LLAMA_SWA_TYPE_NONE.

其他1 个来源
来源与时间线
llama.cppb10361
Google Gemini Blog1 个来源

More than 1 billion people are using the Gemini app every month.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/1BSS.max-600x600.format-webp.webp">The Gemini app has officially surpassed 1 billion monthly users, making it the fastest-growing product in Google’s history. Here’s some data about how people are using G…

其他1 个来源
来源与时间线
Google Gemini BlogMore than 1 billion people are using the Gemini app every month.

8月11日周二

Ollama1 个来源

v0.32.9

NVIDIA Nemotron 3.5 Lightning NVIDIA Nemotron 3.5 Lightning is an open 30B mixture-of-experts (MoE) model with 3B active parameters built for that execution layer of always-on agents. It is designed for harnesses like OpenClaw and Hermes Agent – all supported by the NVIDIA NemoClaw open source security and management stack for running always-on AI agents. ``` ollama run nemotron-3.5-lightning ``` What's Changed * Added the Nemotron 3 architecture * Handle boundary condition in Muse Glimmer function calling parser **Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.8...v0.32.9

其他1 个来源
来源与时间线
Ollamav0.32.9
llama.cpp1 个来源

b10360

<details open> common/peg : suppress incomplete escape sequences (#26780) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10

其他1 个来源
来源与时间线
llama.cppb10360
NVIDIA Generative AI Technical Blog1 个来源

NVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents

Long-running AI agents spend most of their time on high-volume execution: tool calls, result validation, and subagent delegation. Using a frontier reasoning...

其他1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA Nemotron 3.5 Lightning Delivers Fast, Accurate Specialized Task Execution for Long-Running Agents
llama.cpp1 个来源

b10358

<details open> Address review comment of PR 25532 (#26852) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10358/llama-b1035

其他1 个来源
来源与时间线
llama.cppb10358
llama.cpp1 个来源

b10357

<details open> opencl: transpose the K tile in local memory for FA prefill kernels (#26428) </details> **Website:** - <https://llama.app> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/rel

其他1 个来源
来源与时间线
llama.cppb10357
arXiv cs.LG1 个来源

Data-Driven Fire-Zone Segmentation for Improved Short-Term Wildfire Prediction

该论文提出一种数据驱动的火区分割方法，用于改进短期野火预测。作者使用无监督分割算法，结合分水岭检测和K-means聚类，直接从历史火灾模式中定义预测单元，替代传统的均匀网格。在法国六个省和六种预测模型上的实验表明，火区分割在平均IoU上比基于网格的方法提升3–6%（取决于空间尺度），且计算轻量（每次配置<10秒），可完全并行化。

研究1 个来源
来源与时间线
arXiv cs.LGData-Driven Fire-Zone Segmentation for Improved Short-Term Wildfire Prediction
arXiv cs.CV1 个来源

PragyaDoc: A Universal Document Intelligence Framework for Multilingual Medical Document Understanding in Low-Resource Settings

本文介绍了PragyaDoc，一个面向低资源环境下多语言医疗文档理解的通用文档智能框架。它旨在解决印度因22种官方语言导致的医疗文档可及性障碍，其中大多数医疗文档为英文，而农村居民、ASHA工作者和患者家属等最需要这些信息的人群却无法理解。该框架采用四层流水线：并行集成OCR提取层、几何-词汇融合层、确定性领域结构化层和双LLM医疗推理与定位层。

研究1 个来源
来源与时间线
arXiv cs.CVPragyaDoc: A Universal Document Intelligence Framework for Multilingual Medical Document Understanding in Low-Resource Settings
arXiv cs.AI1 个来源

Determinization in Structure Theories: A Unified Framework via Closure, Comparability, and Joint Admissibility

本文提出了一个在多元结构理论中构建规范性解释的统一框架。结构理论被定义为三元组（签名、公理、推理策略），其可接受解释族包含所有全局一致的结构结论指派。作者区分了三个层次的规范化：闭包稳定化、全局完成和确定性化，并将非确定性分为认知多元性（E型）和结构多元性（S型），其中S-强子类以缺乏公共上界为特征。研究提出了基于算子的完成和基于选择器的构造两种规范化机制，给出了它们存在的充分条件，并证明在正非收缩规则及额外可靠性条件下，纯基于推理的完成可归结为饱和闭包算子。对于E型理论，建立了闭包稳定化，而完全确定性化依赖于一个仍开放的整体汇合性质；对于S-强理论，通过规范性选择实现确定性化。此外，多级规范化通过分段算子形成结构上非交换的系统，并提供了一个条件分类定理，将理论内在机制归结为闭包或选择。该框架也适用于LLM辅助推理，其中幻觉可视为不受支持的规范化。

研究1 个来源
来源与时间线
arXiv cs.AIDeterminization in Structure Theories: A Unified Framework via Closure, Comparability, and Joint Admissibility
arXiv cs.CV1 个来源

NeuroPilot: An Agent-Driven Smart Pipeline for Processing, Quality Control, and Managing Neuroimages

NeuroPilot 是一个由 LLM 驱动的多代理智能流水线，用于神经影像处理、质量控制和数据管理。它将神经影像处理、QC 和数据管理的专业知识数字化为三个可调用的技能：dcm2bids-skill、neuroimage-pre-skill 和 qc-agent-skill。该系统在 17 个队列（超过 123,000 名受试者）中展示出通用性，涵盖从婴儿到老年的多种 MRI 模态。QC 代理筛查了 558 名生产受试者，并通过 FreeSurfer 拓扑缺陷指标验证了其自动化标志。婴儿处理管线在 QC 验证的输入上实现了 100%（201/201）的完成率，并将传统 2-3 个月的处理时间缩短至一周。

智能体1 个来源
来源与时间线
arXiv cs.CVNeuroPilot: An Agent-Driven Smart Pipeline for Processing, Quality Control, and Managing Neuroimages
arXiv cs.AI1 个来源

Towards an Argumentative Foundation for Evaluative AI

本文为评估型人工智能（Evaluative AI, EAI）提出论证基础。EAI 不提供单一推荐，而是通过呈现竞争性假设及其正反证据来支持人类决策。作者主张计算论证是该类可解释、可质疑的 EAI 的合适形式化基础，并为分布式和人本 EAI 系统的长期研究议程奠定基础。

研究1 个来源
来源与时间线
arXiv cs.AITowards an Argumentative Foundation for Evaluative AI
arXiv cs.LG1 个来源

Application of Artificial Intelligence for Fraudulent Banking Operations Recognition

本文研究人工智能在银行欺诈交易识别中的应用。鉴于新冠疫情推动银行业务线上化，欺诈案件增多，研究聚焦机器学习算法分析在线银行交易。其科学新颖性在于开发识别欺诈交易的机器学习模型及银行数据预处理技术，并探讨处理不平衡数据集、特征转换和特征工程等提升精度的方法。基于人工神经网络的模型有效提升了欺诈检测准确率；可视化比较显示逻辑回归表现最佳（AUC约0.946），堆叠泛化达到更高AUC 0.954。研究认为AI识别银行欺诈在数字社会具有现实意义。

监管1 个来源
来源与时间线
arXiv cs.LGApplication of Artificial Intelligence for Fraudulent Banking Operations Recognition
arXiv cs.CL1 个来源

DocAtlas: Long-Document Understanding as Mutable-State Interaction

DocAtlas：将长文档理解视为可变状态交互。该系统提出了一种可变文档 harness，通过外部环境控制搜索、读取、记录和回顾文档信息，结合自改进检索、选择性证据访问和主动工作记忆，在固定上下文预算下工作。使用 GPT-5.4 时，DocAtlas 在 MMLongBench-Doc 上达到 71.4% 的准确率，超过人类专家参考值 65.8%；通过端到端强化学习训练的 Qwen3.5-4B 小型 VLM 达到 63.7%，显著超过直接输入基线 54.4%。

智能体1 个来源
来源与时间线
arXiv cs.CLDocAtlas: Long-Document Understanding as Mutable-State Interaction
arXiv cs.CL1 个来源

WuYuEval: A Multi-Level Benchmark for Large Language Models in Solid Waste Management

WuYuEval：面向固体废物管理的大语言模型多层级基准。该论文提出一个评估LLM在固体废物管理领域能力的新基准，包含基础知识与专家决策两个模块，并基于33个模型的评测发现推理模式带来的提升并非普遍积极。

基准1 个来源
来源与时间线
arXiv cs.CLWuYuEval: A Multi-Level Benchmark for Large Language Models in Solid Waste Management
arXiv cs.AI1 个来源

Flow-by-Flow:Content-Judgment Bypass for Governing AI Output in High-Loss Domains

这篇论文提出“Flow-by-Flow”治理范式，旨在不评估内容的情况下控制AI输出的监督负载。作者认为高损失领域中人类监督受限于V×L（输出速度×认知负载），其中分类、判断和响应成本不对称变化。他们提出基于形式化可计数特征的认知成本评分和机构容量上限，并推导出四个设计不变量。蒙特卡洛分析显示复合多指标流控制在90.8%试验中优于单纯监督强化。

研究1 个来源
来源与时间线
arXiv cs.AIFlow-by-Flow:Content-Judgment Bypass for Governing AI Output in High-Loss Domains
arXiv cs.CV1 个来源

Performance of large language models in the optical diagnosis of colorectal polyps

该研究评估了多种多模态大语言模型（MLLM）在结直肠息肉光学诊断中的表现，使用PRIME数据集（132例白光和窄带成像图像）。模型包括Claude Opus 4、Google Gemini 2.5 Pro、GPT-o3、GPT-4o和GPT-5。结果显示，所有模型在区分肿瘤性与非肿瘤性息肉的F1分数均>0.9；Gemini 2.5 Pro在区分浸润性与非浸润性息肉及低级别与高级别腺瘤方面F1分数最高（分别为0.560和0.492）；Claude Opus 4和GPT-5在巴黎分型上的正确率显著高于其他模型（41.7%）。结论认为这些模型在区分息肉亚型上接近专家共识，但敏感性和特异性未达到ESGE标准，需进一步前瞻性多中心试验及人机协同工作流设计。

研究1 个来源
来源与时间线
arXiv cs.CVPerformance of large language models in the optical diagnosis of colorectal polyps
llama.cpp1 个来源

b10356

<details open> ci : target ROCm 7.14 for build and release (#25775) * Switch ROCm from 7.2.1 to 7.14 ROCm 7.14 is the first production release using TheRock build system. It can be installed using multi-arch deliverables from wheels, debs, rpms, tarballs or runfiles. Adjust ROCm targets for Linux and Windows to use this instead. * ci: switch all other Windows ROCm jobs to ROCm 7.14 wheels Move the shared windows-setup-rocm composite action from the HIP SDK PRO Edition installer to the multi-arch ROCm wheels (rocm[libraries,devel]). The wheel-install logic that previously lived inline in release.yml is now in the shared action, and both build-cache.yml and release.yml call it. Also migrate th

其他1 个来源
来源与时间线
llama.cppb10356
Ollama1 个来源

v0.32.8

Muse Glimmer Muse Glimmer is now available on all platforms. Muse Glimmer can power coding agent applications such as Claude Code, Codex, Pi and more, as well as long-running personal assistants such as OpenClaw and Hermes. Ollama's MLX engine provides state-of-the-art performance on Apple Silicon for this model, with support for DFlash and image input as of Ollama 0.32.7. To download and run Muse Glimmer locally: ```bash ollama run muse-glimmer ``` To run Muse Glimmer with Claude Code, download Ollama and run: ```bash ollama launch claude --model muse-glimmer ``` For a lighter-weight coding agent, try Pi: ```bash ollama launch pi --model m

其他1 个来源
来源与时间线
Ollamav0.32.8
The Verge AI1 个来源

Mark Zuckerberg doesn’t understand how to live

The Verge的评论文章，标题批评马克·扎克伯格“不懂如何生活”。文章开头讲述作者遇到一位攀岩者，他用AI制作了一张励志海报（一只熊在峡谷上走扁带，举着写着“做酷事”的牌子），作者对此反应礼貌而保留，暗示这种AI生成的激励文化与扎克伯格对AI和生活的理解形成对比。

观点1 个来源
来源与时间线
The Verge AIMark Zuckerberg doesn’t understand how to live
vLLM1 个来源

v0.27.0

vLLM v0.27.0 发布，包含 561 次提交，来自 242 位贡献者（64 位新贡献者）。主要亮点：完整支持 Kimi K3（核心模型、Python/Rust 前端、AttnRes 内核、DeepGEMM、压缩张量量化检查点、DSpark AR 融合等）；新增 Qwen3.5（文本密集型和 MoE 模型，含 EVS 视频 token 剪枝）、K-EXAONE-2.0-750B-A37B、VaultGemma、jina-embeddings-v5-text-nano 等模型；升级至 PyTorch 2.13.0（伴随 torchvision 0.28.0 和 Triton 3.7.1，属破坏性环境变更，XPU/CPU 也跟进）；深化 FlashAttention 4 在 SM100 上的集成，支持 FP8 KV 缓存和 headdim-256，并通过 JIT 预热消除首次请求编译延迟；同时提到 DeepSeek-V4 性能改进（原文截断，细节未完整）。

基础设施1 个来源
来源与时间线
vLLMv0.27.0
MIT Technology Review AI1 个来源

AI professors are negotiating the new realities of academic research

《麻省理工科技评论》AI栏目报道，AI教授们正在应对学术研究的新现实。文章开头提到作者前往加州山景城参加一场汇聚资深与有潜力的AI研究者的活动，但正文内容截断，仅此片段可见。

研究1 个来源
来源与时间线
MIT Technology Review AIAI professors are negotiating the new realities of academic research
Anthropic Research1 个来源

Learning more about Claude's mathematical capabilities

Anthropic Research 于2026年8月10日发布文章《进一步了解Claude的数学能力》，介绍一个未发布版本的Claude在黎曼猜想（Riemann hypothesis）相关问题上的进展：它将满足黎曼猜想的黎曼ζ函数零点比例的下界从41.6%提高到67.2%。文章称，Anthropic的两位数学家研究并验证了Claude的论文，Claude还生成了可正式验证的证明；外部专家Brian Conrey和Dan Goldston也对该论文进行了审查。

研究1 个来源
来源与时间线
Anthropic ResearchLearning more about Claude's mathematical capabilities

8月10日周一

Ollama1 个来源

v0.32.7

Muse Glimmer > Note: Muse Glimmer is currently available via initial support via Ollama's MLX engine on Apple Silicon. Additional support and optimizations for Apple Silicon, NVIDIA, AMD, and other platforms will be available in the coming days. **Muse Glimmer**, Meta's newest open model and the first released by Meta Superintelligence Labs, is now available on Ollama. It's a 30B multimodal model purpose-built for agent workloads that run locally. With Ollama, you can now use Muse Glimmer to power coding agent applications such as Claude Code, Codex, Pi and more, as well as long-running personal assistants such as OpenClaw and Hermes. Ollama's MLX engine provides state-of-the-art performa

其他1 个来源
来源与时间线
Ollamav0.32.7
Hugging Face Transformers1 个来源

Release: v5.15.0

Hugging Face Transformers 发布 v5.15.0，新增模型支持：Meta Muse Glimmer（专为智能体场景设计的多模态模型，由 Muse 蒸馏至 30B 参数，Apache 2.0 许可，含 2B ViT 视觉编码器和 28B 文本解码器，可本地部署）以及 GraniteMoeSWA 与 GraniteSWA 模型。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease: v5.15.0
MIT Technology Review AI1 个来源

AI for science needs reasoning, not just data

文章标题为“AI用于科学需要推理，而不仅仅是数据”。文章开头提及历史上关于科学终结的预测，如1903年迈克尔逊声称物理事实均已发现，以及20世纪80年代霍金预测理论物理学将在世纪末完成，进而引出人工智能对科学的影响。基于现有摘要，文章主张AI应用于科学需要推理能力，而非仅依赖数据。

观点1 个来源
来源与时间线
MIT Technology Review AIAI for science needs reasoning, not just data
arXiv cs.LG1 个来源

Sharding Prevents LLM Oversight Failures and Adversarial Exploitation

论文《Sharding Prevents LLM Oversight Failures and Adversarial Exploitation》（意为：分片可防止LLM监督失败与对抗性利用）指出，给LLM法官更多计算量并不一定能使其检查更多要求；当一次调用需要返回多个判定时，与专家的一致性会下降。通过将要求分成小组、每组单独调用并汇总结果的分片（sharding）方法，可以在固定预算下提高一致性，且分片的较弱法官可胜过更强大的整体法官。分片对仅改变呈现方式的对抗性攻击具有鲁棒性，但无法应对逐条说服式攻击；在此基础上增加辩论式反对（debate-style opposition）可以抵御此类自适应攻击。

安全1 个来源
来源与时间线
arXiv cs.LGSharding Prevents LLM Oversight Failures and Adversarial Exploitation
arXiv cs.CL1 个来源

TEXAS: Task-Expert-Aware Supervision for Downstream Mixture-of-Experts LLM Adaptation

本文提出了一种名为TEXAS（任务专家感知监督）的方法，用于混合专家（MoE）语言模型的下游适应。该方法通过比较基座模型成功与失败实例上的专家激活情况，保留在成功实例上激活更强的专家；在微调时，对失败实例中激活这些专家的答案标记赋予更高权重。在三个MoE模型和六个基准测试上，TEXAS在18个设置中达到最佳或并列最佳17项，平均比最强基线提升1.3–1.5个百分点。

研究1 个来源
来源与时间线
arXiv cs.CLTEXAS: Task-Expert-Aware Supervision for Downstream Mixture-of-Experts LLM Adaptation
arXiv cs.AI1 个来源

Towards Multi-Label Graph Foundation Models: from Single-Vector Representation Learning to Multi-Semantic Basis Learning

本文提出了一种面向多标签节点分类的图基础模型（MSB-GFM），旨在解决现有图基础模型基于单标签假设导致多语义纠缠和跨域泛化受限的问题。模型采用多语义基元表示学习，将每个多标签节点建模为语义基元的自适应组合，并设计语义-结构双通道架构与域对抗训练，以实现跨域知识迁移。实验验证了模型的有效性。

研究1 个来源
来源与时间线
arXiv cs.AITowards Multi-Label Graph Foundation Models: from Single-Vector Representation Learning to Multi-Semantic Basis Learning
arXiv cs.CV1 个来源

TransSLR: A Lightweight Transformer for Sign Language Recognition

本文提出 TransSLR，一种轻量级Transformer，用于手语识别，特别是中非手语（CASL）。在CASL-W60基准上达到80.39%的准确率，超越此前最优结果+10.46%。该模型使用64帧归一化姿态序列训练，无需依赖RGB外观，适合资源受限环境。

研究1 个来源
来源与时间线
arXiv cs.CVTransSLR: A Lightweight Transformer for Sign Language Recognition
arXiv cs.AI1 个来源

Beyond Routing Weights: Faithful Response-Level Interpretation of Mixture-of-Experts Reward Models via Contribution Contrast

公开信息显示，该条目聚焦“Beyond Routing Weights: Faithful Response-Level Interpretation of Mixture-of-Experts Reward Models via Contribution Contrast”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.AIBeyond Routing Weights: Faithful Response-Level Interpretation of Mixture-of-Experts Reward Models via Contribution Contrast
arXiv cs.CV1 个来源

Deep Evidential Regression for Sparse Forest Height Estimation from Multimodal Satellite Imagery

本文研究深度证据回归（Deep Evidential Regression）在基于多模态卫星影像的稀疏森林高度估计中的应用，在TreeUQ基准上使用Sentinel-1/2数据，提出掩膜证据损失以应对标签稀疏，实现与确定性U-Net相当的预测性能并额外提供校准的不确定性估计。

研究1 个来源
来源与时间线
arXiv cs.CVDeep Evidential Regression for Sparse Forest Height Estimation from Multimodal Satellite Imagery
arXiv cs.LG1 个来源

Risk-Aware Decision Policies for Agents Under Noisy Perception

本文来自 arXiv cs.LG，题为《风险感知的噪声感知决策策略》。文章提出了一种人工生命捕食者-猎物模型，研究噪声感知下的觅食行为，并比较了考虑噪声预测的各种决策策略。实验表明，盲目信任感知标签会导致灾难性失败，而不确定性感知策略能显著提高生存率并减少致命错误，尤其在非对称噪声下。随着不确定性增加，代理行为从探索性转向保守性。该研究将风险敏感觅食、生态信息使用与对噪声标签的鲁棒学习联系起来。

研究1 个来源
来源与时间线
arXiv cs.LGRisk-Aware Decision Policies for Agents Under Noisy Perception
Hugging Face Blog1 个来源

Meta is back with Muse Glimmer: local, agentic, multimodal, and open source

Hugging Face 官方博客发布了题为“Meta 携 Muse Glimmer 回归：本地、智能体、多模态、开源”的文章。该标题表明 Meta 发布了名为 Muse Glimmer 的模型，具有本地运行、智能体、多模态和开源的特点。但本次仅获取到元数据和标题，未包含正文内容。

开源1 个来源
来源与时间线
Hugging Face BlogMeta is back with Muse Glimmer: local, agentic, multimodal, and open source

8月9日周日

llama.cpp1 个来源

b10333

公开信息显示，该条目聚焦“b10333”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10333
llama.cpp1 个来源

b10332

公开信息显示，该条目聚焦“b10332”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10332
The Verge AI1 个来源

An Amazon data center could have the worst polluting power plant in the country

公开信息显示，该条目聚焦“An Amazon data center could have the worst polluting power plant in the country”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

融资1 个来源
来源与时间线
The Verge AIAn Amazon data center could have the worst polluting power plant in the country

8月8日周六

SGLang1 个来源

v0.5.17

SGLang v0.5.17 版本亮点：来自194位贡献者的582个PR；对 Kimi K3 的 Day-0 支持（2.8T 多模态 LatentMoE，原生 MXFP4，1M 上下文等优化）；对 MiniMax-H3（视频生成模型）的 Day-0 支持。

基础设施1 个来源
来源与时间线
SGLangv0.5.17
llama.cpp1 个来源

b10326

公开信息显示，该条目聚焦“b10326”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10326
llama.cpp1 个来源

b10322

公开信息显示，该条目聚焦“b10322”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10322
llama.cpp1 个来源

b10321

公开信息显示，该条目聚焦“b10321”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10321
Anthropic Python SDK1 个来源

v0.121.0

Anthropic Python SDK v0.121.0 发布，新增 mid-conversation-tool-changes beta 支持，以及会话预算、advisor 工具、固定推理位置和从 GitHub 自动加载 skills 等功能；同时移除已退役的 Claude Opus 4.1 模型，并更新文档和依赖约束。

产品更新1 个来源
来源与时间线
Anthropic Python SDKv0.121.0

8月7日周五

OGX1 个来源

v1.3.0

OGX v1.3.0 版本发布。本版本主要变更包括：新增 Meta AI 远程推理提供方；为 Meta 提供方 HTTP 客户端注入网络/TLS 配置；提取 Anthropic 兼容提供方共用的 SSE 透传流辅助函数；修复 vLLM Anthropic 消息 URL 中的重复 /v1 前缀；为 OpenAIMixin 增加 anthropic_count_tokens；并修复 CI 与容器启动问题。

产品更新1 个来源
来源与时间线
OGXv1.3.0
arXiv cs.AI1 个来源

Agentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services

公开信息显示，该条目聚焦“Agentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIAgentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services
arXiv cs.CL1 个来源

Universal Pathologies, Conditional Consequences: A Triple-Robustness Analysis of RAG for Multi-Hop Traceability

公开信息显示，该条目聚焦“Universal Pathologies, Conditional Consequences: A Triple-Robustness Analysis of RAG for Multi-Hop Traceability”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLUniversal Pathologies, Conditional Consequences: A Triple-Robustness Analysis of RAG for Multi-Hop Traceability
arXiv cs.CL1 个来源

Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support

公开信息显示，该条目聚焦“Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLSimulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support
Anthropic News1 个来源

Improving Fable 5 Safeguards

Anthropic宣布对Claude Fable 5的生物安全防护进行更新，大幅减少回退（fallbacks）。测试显示，与生物学相关的回退减少了约85%，Fable 5现可处理更多生物学任务，如解读化验结果、理解症状及教育性生物问题。然而，对于双重用途的请求（如病毒学、毒理学和分子设计），Fable 5仍会回退到Opus 5，尚不能用于专业生物学研究或药物开发。公司正通过可信访问途径来缩小这一差距，同时管理模型能力带来的风险。

安全1 个来源
来源与时间线
Anthropic NewsImproving Fable 5 Safeguards

8月6日周四

OpenAI News1 个来源

Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

公开信息显示，该条目聚焦“Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

模型发布1 个来源
来源与时间线
OpenAI NewsImproving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users
Microsoft Semantic Kernel1 个来源

python-1.44.1

Microsoft Semantic Kernel 发布了 Python 1.44.1 版本。主要变更包括：对 OpenAPI 服务器变量值进行编码；合并 Dependabot 依赖更新；抑制内部 HTTP 工具中的 CodeQL 误报；为 Azure AI Agent 添加 MCP 工具审批回调（破坏性变更）；跳过归一化名称冲突的 MCP 工具和提示词；更新 form-data 依赖至 4.0.6；为 Copilot Studio agent 记录 x5t 证书指纹哈希；并将 Python 版本提升至 1.44.1。

开源1 个来源
来源与时间线
Microsoft Semantic Kernelpython-1.44.1
Microsoft Semantic Kernel1 个来源

dotnet-1.79.0

Microsoft Semantic Kernel 发布 .NET 1.79.0 版本，主要变更包括：包版本更新至 1.79.0、.NET SDK 升级至 10.0.302、依赖项 form-data 更新至 4.0.6、Microsoft.AspNetCore.OData 更新至 9.5.0、修复 CosmosDB vectorstore 错误、将 VectorStoreRAG 和 Concepts 示例迁移至 CommunityToolkit.VectorData 包、为 TimePlugin 添加 TimeProvider 注入以支持确定性测试，以及多项内部工作流与依赖改进。

开源1 个来源
来源与时间线
Microsoft Semantic Kerneldotnet-1.79.0
OpenAI News1 个来源

Working with the American Psychological Association on youth mental health and AI

公开信息显示，该条目聚焦“Working with the American Psychological Association on youth mental health and AI”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
OpenAI NewsWorking with the American Psychological Association on youth mental health and AI
arXiv cs.AI1 个来源

The LLM Proposes, the Executive Disposes: A Self-Verifying Agent Instrument that Dissociates Commitment Drift from Binding Drift in Long-Horizon Agents

公开信息显示，该条目聚焦“The LLM Proposes, the Executive Disposes: A Self-Verifying Agent Instrument that Dissociates Commitment Drift from Binding Drift in Long-Horizon Agents”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIThe LLM Proposes, the Executive Disposes: A Self-Verifying Agent Instrument that Dissociates Commitment Drift from Binding Drift in Long-Horizon Agents
arXiv cs.AI1 个来源

A Long-Run Persistence Theory for AI Systems under the Redundancy-Adjusted Artificial Age Score (AAS)

公开信息显示，该条目聚焦“A Long-Run Persistence Theory for AI Systems under the Redundancy-Adjusted Artificial Age Score (AAS)”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

融资1 个来源
来源与时间线
arXiv cs.AIA Long-Run Persistence Theory for AI Systems under the Redundancy-Adjusted Artificial Age Score (AAS)
arXiv cs.CV1 个来源

LoRetta: A Foundation Model and Extensive Dataset for Global-Scale Remote Sensing Dense Image Matching

公开信息显示，该条目聚焦“LoRetta: A Foundation Model and Extensive Dataset for Global-Scale Remote Sensing Dense Image Matching”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CVLoRetta: A Foundation Model and Extensive Dataset for Global-Scale Remote Sensing Dense Image Matching
arXiv cs.LG1 个来源

C$^2$MOE: Consistency and Complementarity-guided Mixture of Experts for Incomplete Multimodal Emotion Learning

公开信息显示，该条目聚焦“C$^2$MOE: Consistency and Complementarity-guided Mixture of Experts for Incomplete Multimodal Emotion Learning”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.LGC$^2$MOE: Consistency and Complementarity-guided Mixture of Experts for Incomplete Multimodal Emotion Learning
arXiv cs.CL1 个来源

MemArena: An Ego-Centric Benchmark for On-Device Agentic Personal Memory Assistants at Scale

公开信息显示，该条目聚焦“MemArena: An Ego-Centric Benchmark for On-Device Agentic Personal Memory Assistants at Scale”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.CLMemArena: An Ego-Centric Benchmark for On-Device Agentic Personal Memory Assistants at Scale
OpenAI News1 个来源

From asking to doing: How the world is putting ChatGPT to work

公开信息显示，该条目聚焦“From asking to doing: How the world is putting ChatGPT to work”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
OpenAI NewsFrom asking to doing: How the world is putting ChatGPT to workOpenAI NewsNew ways to learn and teach with ChatGPT Work and Codex
Microsoft Foundry Blog1 个来源

Exploring Multi-Agent Workflows with Microsoft Agent Framework | Microsoft Community Hub

该原始条目是微软 Foundry 博客在 Microsoft Community Hub 上发布的一篇文章，标题为“Exploring Multi-Agent Workflows with Microsoft Agent Framework”，发布于 2026 年 8 月 5 日。根据抓取到的开头片段，文章内容围绕组织从孤立的 AI 代理转向多代理工作流，以扩展决策、内容生成、分析和执行的自动化。由于正文被截断，仅基于标题和开头片段总结，未获取完整文章内容。

智能体1 个来源
来源与时间线
Microsoft Foundry BlogExploring Multi-Agent Workflows with Microsoft Agent Framework | Microsoft Community Hub

8月5日周三

OpenAI Cookbook1 个来源

Add an evidence-grounded AML Agents SDK example for Amazon Bedrock (#2896)

Add an evidence-grounded AML Agents SDK example for Amazon Bedrock (#2896) Adds a beginner-friendly AML Agents SDK example with evidence-grounded outputs, deterministic validation, human-review boundaries, and Amazon Bedrock integration.

其他1 个来源
来源与时间线
OpenAI CookbookAdd an evidence-grounded AML Agents SDK example for Amazon Bedrock (#2896)
arXiv cs.AI1 个来源

Enhancing LLMs with Context-Specific Knowledge for Mitigating Misinformation in SMEs: A RAG-based Modeling and Analysis

公开信息显示，该条目聚焦“Enhancing LLMs with Context-Specific Knowledge for Mitigating Misinformation in SMEs: A RAG-based Modeling and Analysis”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.AIEnhancing LLMs with Context-Specific Knowledge for Mitigating Misinformation in SMEs: A RAG-based Modeling and Analysis
arXiv cs.AI1 个来源

Revisiting Classic Thought Experiments to Measure Consciousness for Artificial Intelligence Safety

公开信息显示，该条目聚焦“Revisiting Classic Thought Experiments to Measure Consciousness for Artificial Intelligence Safety”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.AIRevisiting Classic Thought Experiments to Measure Consciousness for Artificial Intelligence Safety
Ars Technica AI1 个来源

Texas halts data center connections to power grid amid overwhelming demand

公开信息显示，该条目聚焦“Texas halts data center connections to power grid amid overwhelming demand”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AITexas halts data center connections to power grid amid overwhelming demand
Ollama1 个来源

v0.32.6

公开信息显示，该条目聚焦“v0.32.6”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
Ollamav0.32.6
Anthropic News1 个来源

Tino Cuellar joins Anthropic as Chief Global Affairs Officer

Anthropic 宣布 Mariano-Florentino (Tino) Cuéllar 将加入公司，担任首任首席全球事务官 (Chief Global Affairs Officer)，负责政策、国际战略参与及全球政府关系。Cuéllar 曾担任卡内基国际和平基金会主席、加州最高法院法官、斯坦福大学弗里曼·斯波利国际研究所所长等职，并自 2026 年 1 月起担任 Anthropic 长期利益信托的受托人。他将卸任受托人职务加入公司，信托将按正常程序遴选继任者。

监管1 个来源
来源与时间线
Anthropic NewsTino Cuellar joins Anthropic as Chief Global Affairs Officer

8月4日周二

arXiv cs.CL1 个来源

RubricReviewer: From Direct Critique to Objective and Comprehensive Rubric-Driven Peer Review

公开信息显示，该条目聚焦“RubricReviewer: From Direct Critique to Objective and Comprehensive Rubric-Driven Peer Review”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.CLRubricReviewer: From Direct Critique to Objective and Comprehensive Rubric-Driven Peer Review
llama.cpp1 个来源

b10251

公开信息显示，该条目聚焦“b10251”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10251
llama.cpp1 个来源

b10250

公开信息显示，该条目聚焦“b10250”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10250
llama.cpp1 个来源

b10249

公开信息显示，该条目聚焦“b10249”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10249
llama.cpp1 个来源

b10245

公开信息显示，该条目聚焦“b10245”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10245
llama.cpp1 个来源

b10244

公开信息显示，该条目聚焦“b10244”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10244
OpenAI Python SDK1 个来源

v2.53.0

OpenAI Python SDK v2.53.0 发布（2026-08-03）。主要更新：在 Responses 类型中新增对 gpt-5.5 模型以及工具名称/命名空间的支持；修复 CI 中避免 NumPy 源码构建和重复 HTTPX 覆盖率的问题。

监管1 个来源
来源与时间线
OpenAI Python SDKv2.53.0
llama.cpp1 个来源

b10243

公开信息显示，该条目聚焦“b10243”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10243
NVIDIA Generative AI Technical Blog1 个来源

NVIDIA Vera Storage Benchmarks: Faster Encryption, Compression, Integrity Checking, and Recovery for AI-Native Storage

NVIDIA 官方博客发布关于“NVIDIA Vera 存储基准测试”的内容，主题为更快的加密、压缩、完整性检查和恢复能力，面向 AI 原生存储。文章标题和摘要指出，存储是每个 agentic AI 工作流中的活跃部分，涉及智能体检索企业知识、访问持久内存、复用键值（KV）缓存数据等场景。

基准1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA Vera Storage Benchmarks: Faster Encryption, Compression, Integrity Checking, and Recovery for AI-Native Storage

8月3日周一

arXiv cs.CL1 个来源

Can LLMs Really Understand Item Difficulty Levels? Implications for Automated Item Generation Using LLMs

公开信息显示，该条目聚焦“Can LLMs Really Understand Item Difficulty Levels? Implications for Automated Item Generation Using LLMs”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLCan LLMs Really Understand Item Difficulty Levels? Implications for Automated Item Generation Using LLMs
arXiv cs.AI1 个来源

OpenClaw and Ollama in Agentic AI: Toward Fully Autonomous and Scalable AI Agent Systems

公开信息显示，该条目聚焦“OpenClaw and Ollama in Agentic AI: Toward Fully Autonomous and Scalable AI Agent Systems”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIOpenClaw and Ollama in Agentic AI: Toward Fully Autonomous and Scalable AI Agent Systems
arXiv cs.AI1 个来源

Can AI Evaluate AI Scientists? A Benchmarking Study of Autonomous Research Generation Systems Using Automated Multi-Model Review

公开信息显示，该条目聚焦“Can AI Evaluate AI Scientists? A Benchmarking Study of Autonomous Research Generation Systems Using Automated Multi-Model Review”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.AICan AI Evaluate AI Scientists? A Benchmarking Study of Autonomous Research Generation Systems Using Automated Multi-Model Review
arXiv cs.AI1 个来源

LLM Framework for Discovering Major Mathematical Conjectures: AI's Quest for the Next Riemann Hypothesis

公开信息显示，该条目聚焦“LLM Framework for Discovering Major Mathematical Conjectures: AI's Quest for the Next Riemann Hypothesis”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.AILLM Framework for Discovering Major Mathematical Conjectures: AI's Quest for the Next Riemann Hypothesis

8月2日周日

8月1日周六

NVIDIA Generative AI Technical Blog1 个来源

Co-Designing AI Model Attention for Fast, Interactive Long-Context Inference

公开信息显示，该条目聚焦“Co-Designing AI Model Attention for Fast, Interactive Long-Context Inference”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogCo-Designing AI Model Attention for Fast, Interactive Long-Context Inference
Ars Technica AI1 个来源

Google Earth risked ruin with retracted AI tool for making fake satellite pics

公开信息显示，该条目聚焦“Google Earth risked ruin with retracted AI tool for making fake satellite pics”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AIGoogle Earth risked ruin with retracted AI tool for making fake satellite pics

7月31日周五

OpenAI Python SDK1 个来源

v2.52.0

OpenAI Python SDK 发布 v2.52.0 版本。主要更新：新增 API 内容来源（provenance）检查功能；修复客户端最多等待两分钟 Retry-After 延迟的问题；新增 API-key mTLS HTTP 客户端配置文档。

产品更新1 个来源
来源与时间线
OpenAI Python SDKv2.52.0
arXiv cs.AI1 个来源

ClinLens: Towards Long-Horizon Coding Agents for Longitudinal Multimodal Clinical Data Science

公开信息显示，该条目聚焦“ClinLens: Towards Long-Horizon Coding Agents for Longitudinal Multimodal Clinical Data Science”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIClinLens: Towards Long-Horizon Coding Agents for Longitudinal Multimodal Clinical Data Science
arXiv cs.AI1 个来源

Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems

公开信息显示，该条目聚焦“Even More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIEven More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent Systems
arXiv cs.CL1 个来源

AI-assisted pre-review of open-source software submissions: an experience report from BOSC 2026

公开信息显示，该条目聚焦“AI-assisted pre-review of open-source software submissions: an experience report from BOSC 2026”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
arXiv cs.CLAI-assisted pre-review of open-source software submissions: an experience report from BOSC 2026
arXiv cs.AI1 个来源

Probing the Origins of Reasoning Performance: Representational Quality for Mathematical Problem-Solving in RL vs. SFT Fine-Tuned Models

公开信息显示，该条目聚焦“Probing the Origins of Reasoning Performance: Representational Quality for Mathematical Problem-Solving in RL vs. SFT Fine-Tuned Models”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
arXiv cs.AIProbing the Origins of Reasoning Performance: Representational Quality for Mathematical Problem-Solving in RL vs. SFT Fine-Tuned Models
arXiv cs.CL1 个来源

Prompt Chaining in Practice: A Case Study in Automated Scholarly Report Generation

公开信息显示，该条目聚焦“Prompt Chaining in Practice: A Case Study in Automated Scholarly Report Generation”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLPrompt Chaining in Practice: A Case Study in Automated Scholarly Report Generation
NVIDIA Generative AI Technical Blog1 个来源

Run High-Performance Core Math at Scale with NVIDIA nvmath-python

公开信息显示，该条目聚焦“Run High-Performance Core Math at Scale with NVIDIA nvmath-python”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogRun High-Performance Core Math at Scale with NVIDIA nvmath-python
llama.cpp1 个来源

b10199

公开信息显示，该条目聚焦“b10199”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10199
llama.cpp1 个来源

b10198

公开信息显示，该条目聚焦“b10198”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10198
llama.cpp1 个来源

b10197

公开信息显示，该条目聚焦“b10197”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10197
Ars Technica AI1 个来源

Google reveals Gemini Robotics 2.0, promising improved dexterity and safety

公开信息显示，该条目聚焦“Google reveals Gemini Robotics 2.0, promising improved dexterity and safety”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AIGoogle reveals Gemini Robotics 2.0, promising improved dexterity and safety
OpenAI Python SDK1 个来源

v2.51.0

公开信息显示，该条目聚焦“v2.51.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.51.0

7月30日周四

Google DeepMind Blog1 个来源

Gemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration

Google DeepMind 发布了关于 Gemini Robotics ER 2 的博客，介绍其通过视频理解、任务编排和多机器人协作来增强机器人推理与协作能力，以解决现实世界任务。

产品更新1 个来源
来源与时间线
Google DeepMind BlogGemini Robotics ER 2: powering robotics with video understanding, task orchestration, and multi-robot collaboration
OpenAI News1 个来源

How GPT-5.6 fuses frontier intelligence with frontier efficiency

公开信息显示，该条目聚焦“How GPT-5.6 fuses frontier intelligence with frontier efficiency”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
OpenAI NewsHow GPT-5.6 fuses frontier intelligence with frontier efficiencyOpenAI NewsAdvancing the price-performance frontier with GPT-5.6
TechCrunch AI1 个来源

Mark Zuckerberg predicts that billions of people will have personal AI agents in five years

公开信息显示，该条目聚焦“Mark Zuckerberg predicts that billions of people will have personal AI agents in five years”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
TechCrunch AIMark Zuckerberg predicts that billions of people will have personal AI agents in five years
Ars Technica AI1 个来源

Mythos attack on 3rd-round PQC algorithm candidate puts it out of commission

公开信息显示，该条目聚焦“Mythos attack on 3rd-round PQC algorithm candidate puts it out of commission”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AIMythos attack on 3rd-round PQC algorithm candidate puts it out of commission
NVIDIA Generative AI Technical Blog1 个来源

How to Self-Host a Validated AI Coding Assistant with NVIDIA NeMo Guardrails

公开信息显示，该条目聚焦“How to Self-Host a Validated AI Coding Assistant with NVIDIA NeMo Guardrails”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

监管1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogHow to Self-Host a Validated AI Coding Assistant with NVIDIA NeMo Guardrails
Google DeepMind Blog1 个来源

We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control

谷歌DeepMind博客宣布在Google Flow Music中推出Lyria 3.5，其在音乐性、歌词、人声和创意控制方面取得进展。

模型发布1 个来源
来源与时间线
Google DeepMind BlogWe’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control

7月29日周三

Microsoft Foundry Blog1 个来源

Executive AI Avatar Platform on Azure: Frontier Intelligence Meets Fireworks AI Open-Source | Microsoft Community Hub

微软官方博客介绍了一个在Azure上构建的行政AI虚拟人平台，结合前沿智能与Fireworks AI开源技术。文章以开放注册截止前夜员工关于HSA的提问为例，展示AI对常见问题的即时响应能力。来源为Microsoft Community Hub的Azure AI Foundry博客，发布于2026年7月29日。

产品更新1 个来源
来源与时间线
Microsoft Foundry BlogExecutive AI Avatar Platform on Azure: Frontier Intelligence Meets Fireworks AI Open-Source | Microsoft Community Hub
arXiv cs.AI1 个来源

Kernel Forge: An Agent Harness for LLM-based Generation and Optimization of CUDA Kernels

公开信息显示，该条目聚焦“Kernel Forge: An Agent Harness for LLM-based Generation and Optimization of CUDA Kernels”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIKernel Forge: An Agent Harness for LLM-based Generation and Optimization of CUDA Kernels
arXiv cs.AI1 个来源

Beyond Memory: A Templated Substrate for Heterogeneous Collaborative Knowledge Work with LLM Agents

公开信息显示，该条目聚焦“Beyond Memory: A Templated Substrate for Heterogeneous Collaborative Knowledge Work with LLM Agents”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIBeyond Memory: A Templated Substrate for Heterogeneous Collaborative Knowledge Work with LLM Agents
arXiv cs.LG1 个来源

FinAbstain: Uncertainty-Calibrated Multimodal RAG for Selective Financial Forecasting

公开信息显示，该条目聚焦“FinAbstain: Uncertainty-Calibrated Multimodal RAG for Selective Financial Forecasting”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.LGFinAbstain: Uncertainty-Calibrated Multimodal RAG for Selective Financial Forecasting
arXiv cs.CV1 个来源

DisasterTD: Disaster Toponym Disambiguation Using Multimodal LLMs and Cross-View Geolocalization

公开信息显示，该条目聚焦“DisasterTD: Disaster Toponym Disambiguation Using Multimodal LLMs and Cross-View Geolocalization”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CVDisasterTD: Disaster Toponym Disambiguation Using Multimodal LLMs and Cross-View Geolocalization
arXiv cs.CL1 个来源

Neuromorphic Diffusion Language Models: Addressing Compute and Memory Bottlenecks via Sparsity and Block Denoising

公开信息显示，该条目聚焦“Neuromorphic Diffusion Language Models: Addressing Compute and Memory Bottlenecks via Sparsity and Block Denoising”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
arXiv cs.CLNeuromorphic Diffusion Language Models: Addressing Compute and Memory Bottlenecks via Sparsity and Block Denoising
arXiv cs.CL1 个来源

Measuring and Improving Behavioral Consistency in Large Language Models through Fact-Heuristic-Emotion State Enforcement

公开信息显示，该条目聚焦“Measuring and Improving Behavioral Consistency in Large Language Models through Fact-Heuristic-Emotion State Enforcement”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

融资1 个来源
来源与时间线
arXiv cs.CLMeasuring and Improving Behavioral Consistency in Large Language Models through Fact-Heuristic-Emotion State Enforcement
OpenAI Python SDK1 个来源

v2.50.0

公开信息显示，该条目聚焦“v2.50.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.50.0
NVIDIA Generative AI Technical Blog1 个来源

Developing Healthcare Robotics with GPU-Native Medical Physics Simulation

公开信息显示，该条目聚焦“Developing Healthcare Robotics with GPU-Native Medical Physics Simulation”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogDeveloping Healthcare Robotics with GPU-Native Medical Physics Simulation
Ars Technica AI1 个来源

Despite AI hype, Google's data shows workers aren't automating themselves away

公开信息显示，该条目聚焦“Despite AI hype, Google's data shows workers aren't automating themselves away”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AIDespite AI hype, Google's data shows workers aren't automating themselves away

7月28日周二

Hugging Face PEFT1 个来源

v0.20.0

Hugging Face PEFT 发布 v0.20.0 版本，新增九种PEFT方法（包括HiRA和GLoRA），并添加图像生成基准测试、大幅改进文档结构。

开源1 个来源
来源与时间线
Hugging Face PEFTv0.20.0
arXiv cs.AI1 个来源

FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills

公开信息显示，该条目聚焦“FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIFlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills
arXiv cs.AI1 个来源

Risk Is Not the Target: A Monotonic Framework for Evaluating Wildfire Operational Risk Signals

公开信息显示，该条目聚焦“Risk Is Not the Target: A Monotonic Framework for Evaluating Wildfire Operational Risk Signals”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIRisk Is Not the Target: A Monotonic Framework for Evaluating Wildfire Operational Risk Signals
arXiv cs.CL1 个来源

MioFFAn: an Annotation Software for Formula Formalization with LLM Automation Capabilities

公开信息显示，该条目聚焦“MioFFAn: an Annotation Software for Formula Formalization with LLM Automation Capabilities”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLMioFFAn: an Annotation Software for Formula Formalization with LLM Automation Capabilities
arXiv cs.CL1 个来源

Evaluating the Impact of Reviewer Guideline Design on LLM-Based Automated Peer Review

公开信息显示，该条目聚焦“Evaluating the Impact of Reviewer Guideline Design on LLM-Based Automated Peer Review”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLEvaluating the Impact of Reviewer Guideline Design on LLM-Based Automated Peer Review
OpenAI Python SDK1 个来源

v2.49.0

公开信息显示，该条目聚焦“v2.49.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.49.0
MIT Technology Review AI1 个来源

OpenAI called the Hugging Face attack unprecedented. But we’ve been here before.

公开信息显示，该条目聚焦“OpenAI called the Hugging Face attack unprecedented. But we’ve been here before.”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

观点1 个来源
来源与时间线
MIT Technology Review AIOpenAI called the Hugging Face attack unprecedented. But we’ve been here before.
NVIDIA Generative AI Technical Blog1 个来源

NVIDIA Ising Enables Fully Automated Quantum Computer Calibration with Enhanced In-Context Learning

公开信息显示，该条目聚焦“NVIDIA Ising Enables Fully Automated Quantum Computer Calibration with Enhanced In-Context Learning”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA Ising Enables Fully Automated Quantum Computer Calibration with Enhanced In-Context Learning

7月27日周一

OGX1 个来源

v1.2.2

OGX 发布 v1.2.2，包含两个修复：升级 pyasn1 以修复 CVE-2026-59885；在推理模块中为透传 AsyncOpenAI 客户端添加 _enforce_credentials=False。

产品更新1 个来源
来源与时间线
OGXv1.2.2
OGX1 个来源

v0.7.3

OGX v0.7.3（补丁版本）聚焦安全修复：升级 pillow、python-multipart、pyasn1、urllib3、python-dotenv、nltk、langchain-core、aiohttp 和 pyjwt 以修复多个 CVE（如 CVE-2026-40192、CVE-2026-42561 等），并包含一个 CI 修复（将客户端检出固定到对应发布分支）。

安全1 个来源
来源与时间线
OGXv0.7.3
OGX1 个来源

v0.4.7

公开信息显示，该条目聚焦“v0.4.7”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OGXv0.4.7
arXiv cs.CL1 个来源

Adversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-based Stylistic Triggers Optimization

公开信息显示，该条目聚焦“Adversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-based Stylistic Triggers Optimization”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.CLAdversarial Style Optimization: Enhancing VLM Jailbreaks by GRPO-based Stylistic Triggers Optimization
arXiv cs.CL1 个来源

A Consensus-Based Framework for Relative Preference Evaluation of Large Language Models

公开信息显示，该条目聚焦“A Consensus-Based Framework for Relative Preference Evaluation of Large Language Models”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLA Consensus-Based Framework for Relative Preference Evaluation of Large Language Models
arXiv cs.CL1 个来源

Evaluation design conditions the expert-vs-auto MeSH gap: a controlled comparison of bag-of-words and BiomedBERT on the Cohen benchmark

公开信息显示，该条目聚焦“Evaluation design conditions the expert-vs-auto MeSH gap: a controlled comparison of bag-of-words and BiomedBERT on the Cohen benchmark”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLEvaluation design conditions the expert-vs-auto MeSH gap: a controlled comparison of bag-of-words and BiomedBERT on the Cohen benchmark
Ollama1 个来源

v0.32.5

公开信息显示，该条目聚焦“v0.32.5”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
Ollamav0.32.5
vLLM1 个来源

v0.26.0

vLLM v0.26.0 发布，包含 411 次提交，来自 212 位贡献者（其中 61 位新贡献者）。主要亮点：新增 Inkling 模型系列完整支持栈（基础建模、CUDA 图、Hopper FA4 相对注意力、MTP=1 推测解码、LoRA、ModelOpt NVFP4 量化）；针对 DeepSeek-V4 的多厂商性能优化（专用路由内核、fused_topk_bias、冗余 repeat/copy 移除、ROCm 两阶段压缩器、稀疏解码/预填充优化、AMD 和 XPU 上的 DSpark 推测解码）；通过 head_dtype 为生成模型提供 fp32 lm_head，并扩展至 LoRA 路径和 ROCm 快速路径；灵活的注意力后端（可按 KV-cache 组选择后端，滑窗支持成为显式后端能力，改善混合模型支持）。

基础设施1 个来源
来源与时间线
vLLMv0.26.0
llama.cpp1 个来源

b10141

公开信息显示，该条目聚焦“b10141”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10141

7月26日周日

7月25日周六

TechCrunch AI1 个来源

One fallen power line exposed a growing AI data center problem. Here’s how to fix it.

公开信息显示，该条目聚焦“One fallen power line exposed a growing AI data center problem. Here’s how to fix it.”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
TechCrunch AIOne fallen power line exposed a growing AI data center problem. Here’s how to fix it.
arXiv cs.CL1 个来源

Knowledge Injection Exists in MoE? Exploring Expert-Aware Contrast Decoding in MoE for Mitigating LLMs'Hallucinations

公开信息显示，该条目聚焦“Knowledge Injection Exists in MoE? Exploring Expert-Aware Contrast Decoding in MoE for Mitigating LLMs'Hallucinations”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLKnowledge Injection Exists in MoE? Exploring Expert-Aware Contrast Decoding in MoE for Mitigating LLMs'Hallucinations
arXiv cs.CL1 个来源

What is Good? Extracting and Testing Implicit Theories of Literary Quality from LLM Reasoning Traces

公开信息显示，该条目聚焦“What is Good? Extracting and Testing Implicit Theories of Literary Quality from LLM Reasoning Traces”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLWhat is Good? Extracting and Testing Implicit Theories of Literary Quality from LLM Reasoning Traces
Ollama1 个来源

v0.32.4

公开信息显示，该条目聚焦“v0.32.4”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
Ollamav0.32.4
SGLang1 个来源

v0.5.16

SGLang 发布 v0.5.16，包含来自 169 位贡献者的 574 个 PR。主要亮点：1) DSpark 置信度驱动推测解码算法：基于分块半自回归草稿，根据草稿置信度动态调整验证窗口大小；在 DeepSeek-V4-Pro、B300 TP8 配置下达到 383.7 tok/s（接受长度约 5），需设置 --speculative-algorithm DSPARK 和 SGLANG_RAGGED_VERIFY_MODE=compact。2) 支持 Inkling 模型：一个 975B 参数多模态 MoE，拥有 1M token 上下文，融合滑窗注意力、全注意力和 Mamba2 线性注意力，加入 NVFP4 MoE、可选视觉/音频塔及原生 MTP；在 Blackwell 上输入吞吐达 71.7k tok/s，每用户解码 171.0 tok/s，已在 Blackwell TP4/TP8、H200 和 AMD MI350X/MI355X 上验证。

基础设施1 个来源
来源与时间线
SGLangv0.5.16
Ars Technica AI1 个来源

Team uses AlphaFold AI to redesign gene-editing proteins to make them safer

公开信息显示，该条目聚焦“Team uses AlphaFold AI to redesign gene-editing proteins to make them safer”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AITeam uses AlphaFold AI to redesign gene-editing proteins to make them safer
Microsoft Foundry Blog1 个来源

Claude Opus 5 is available today in Microsoft Foundry | Microsoft Community Hub

公开信息显示，该条目聚焦“Claude Opus 5 is available today in Microsoft Foundry | Microsoft Community Hub”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

商业1 个来源
来源与时间线
Microsoft Foundry BlogClaude Opus 5 is available today in Microsoft Foundry | Microsoft Community Hub
Anthropic Research1 个来源

Project Pilot: Can AI models fly drones?

Anthropic与Andon Labs合作推出Drone-Bench基准，测试AI模型能否自主驾驶四旋翼无人机在室内环境中定位并跟随指定人员。该研究基于Project Vend和Project Fetch的进展，强调此类能力的双重用途性质，并指出需要建立有效的规范和治理框架。

智能体1 个来源
来源与时间线
Anthropic ResearchProject Pilot: Can AI models fly drones?
NVIDIA Generative AI Technical Blog1 个来源

ModelExpress: Distributing Model Artifacts at the Speed of Light

公开信息显示，该条目聚焦“ModelExpress: Distributing Model Artifacts at the Speed of Light”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogModelExpress: Distributing Model Artifacts at the Speed of Light

7月24日周五

llama.cpp1 个来源

b10107

公开信息显示，该条目聚焦“b10107”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10107
llama.cpp1 个来源

b10106

公开信息显示，该条目聚焦“b10106”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10106
arXiv cs.AI1 个来源

AINTMA: Agentic AI Architecture for Autonomous Test Management with Generative Intelligence, Secure Cloud Communication and Adaptive Quality Analytics

公开信息显示，该条目聚焦“AINTMA: Agentic AI Architecture for Autonomous Test Management with Generative Intelligence, Secure Cloud Communication and Adaptive Quality Analytics”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIAINTMA: Agentic AI Architecture for Autonomous Test Management with Generative Intelligence, Secure Cloud Communication and Adaptive Quality Analytics
arXiv cs.AI1 个来源

ClickGuard: Detecting and Spoiling Clickbait News with Informativeness Measures and Large Language Models

公开信息显示，该条目聚焦“ClickGuard: Detecting and Spoiling Clickbait News with Informativeness Measures and Large Language Models”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.AIClickGuard: Detecting and Spoiling Clickbait News with Informativeness Measures and Large Language Models
OpenAI Python SDK1 个来源

v2.48.0

公开信息显示，该条目聚焦“v2.48.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.48.0
Anthropic Python SDK1 个来源

v0.119.0

公开信息显示，该条目聚焦“v0.119.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Anthropic Python SDKv0.119.0
Ars Technica AI1 个来源

Google just had its first negative cash flow quarter due to massive AI spending

公开信息显示，该条目聚焦“Google just had its first negative cash flow quarter due to massive AI spending”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

商业1 个来源
来源与时间线
Ars Technica AIGoogle just had its first negative cash flow quarter due to massive AI spending
NVIDIA Generative AI Technical Blog1 个来源

Start Customizing NVIDIA Nemotron 3 Nano with Prime Intellect Lab in Minutes

公开信息显示，该条目聚焦“Start Customizing NVIDIA Nemotron 3 Nano with Prime Intellect Lab in Minutes”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogStart Customizing NVIDIA Nemotron 3 Nano with Prime Intellect Lab in MinutesNVIDIA Generative AI Technical BlogMake Long-Running NVIDIA TensorRT Engine Builds Observable and Cancelable in Python or C++

7月23日周四

arXiv cs.AI1 个来源

BatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data

公开信息显示，该条目聚焦“BatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIBatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data
arXiv cs.CL1 个来源

Stateful Guardrails for Multi-Turn LLM Systems: A Conversational Risk Accumulation Framework

公开信息显示，该条目聚焦“Stateful Guardrails for Multi-Turn LLM Systems: A Conversational Risk Accumulation Framework”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLStateful Guardrails for Multi-Turn LLM Systems: A Conversational Risk Accumulation Framework
Ollama1 个来源

v0.32.3

公开信息显示，该条目聚焦“v0.32.3”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ollamav0.32.3
llama.cpp1 个来源

b10091

公开信息显示，该条目聚焦“b10091”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10091
llama.cpp1 个来源

b10090

公开信息显示，该条目聚焦“b10090”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10090
llama.cpp1 个来源

b10089

公开信息显示，该条目聚焦“b10089”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
llama.cppb10089
OpenAI Python SDK1 个来源

v2.47.0

公开信息显示，该条目聚焦“v2.47.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.47.0
Anthropic News1 个来源

Supporting ambitious external research through the Anthropic Economic Futures Research Fund

公开信息显示，该条目聚焦“Supporting ambitious external research through the Anthropic Economic Futures Research Fund”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
Anthropic NewsSupporting ambitious external research through the Anthropic Economic Futures Research Fund
Anthropic Python SDK1 个来源

v0.118.0

公开信息显示，该条目聚焦“v0.118.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Anthropic Python SDKv0.118.0
Microsoft Foundry Blog1 个来源

Design the Network Before You Deploy: Best Practices for Microsoft Foundry Standard Agents BYOVNet | Microsoft Community Hub

公开信息显示，该条目聚焦“Design the Network Before You Deploy: Best Practices for Microsoft Foundry Standard Agents BYOVNet | Microsoft Community Hub”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Microsoft Foundry BlogDesign the Network Before You Deploy: Best Practices for Microsoft Foundry Standard Agents BYOVNet | Microsoft Community Hub

7月22日周三

Ars Technica AI1 个来源

Unlimited AI tokens aren't unlimited after all as US Army burns through supply

公开信息显示，该条目聚焦“Unlimited AI tokens aren't unlimited after all as US Army burns through supply”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
Ars Technica AIUnlimited AI tokens aren't unlimited after all as US Army burns through supply
arXiv cs.CL1 个来源

A Classifier That Teaches Itself: Self-Improving, Frozen-gate Training (SIFT) for Dynamic Document Classification

公开信息显示，该条目聚焦“A Classifier That Teaches Itself: Self-Improving, Frozen-gate Training (SIFT) for Dynamic Document Classification”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

商业1 个来源
来源与时间线
arXiv cs.CLA Classifier That Teaches Itself: Self-Improving, Frozen-gate Training (SIFT) for Dynamic Document Classification
TechCrunch AI1 个来源

Jack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents

公开信息显示，该条目聚焦“Jack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
TechCrunch AIJack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents
Google Gemini Blog · Ars Technica AI3 个来源

Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

公开信息显示，该条目聚焦“Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

模型发布3 个来源
来源与时间线
Google Gemini BlogIntroducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash CyberArs Technica AIGoogle announces Gemini 3.6 Flash and cybersecurity AI, teases 3.5 Pro and Gemini 4Google DeepMind BlogIntroducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash CyberGoogle DeepMind BlogIntroducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

7月21日周二

NVIDIA Generative AI Technical Blog1 个来源

NVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI

公开信息显示，该条目聚焦“NVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogNVIDIA Vera CPU: Olympus Cores Built for Maximum Single-Thread Performance in Agentic AI
NVIDIA Generative AI Technical Blog1 个来源

Inside NVIDIA Rubin GPU Architecture: Powering the Era of Agentic AI

公开信息显示，该条目聚焦“Inside NVIDIA Rubin GPU Architecture: Powering the Era of Agentic AI”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogInside NVIDIA Rubin GPU Architecture: Powering the Era of Agentic AI
NVIDIA Generative AI Technical Blog1 个来源

Setting a World Record for MoE Pre-Training on NVIDIA GB300 NVL72

公开信息显示，该条目聚焦“Setting a World Record for MoE Pre-Training on NVIDIA GB300 NVL72”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogSetting a World Record for MoE Pre-Training on NVIDIA GB300 NVL72
OpenAI News1 个来源

OpenAI and Hugging Face partner to address security incident during model evaluation

公开信息显示，该条目聚焦“OpenAI and Hugging Face partner to address security incident during model evaluation”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
OpenAI NewsOpenAI and Hugging Face partner to address security incident during model evaluation
arXiv cs.AI1 个来源

Design and Validation of a Lightweight 1D CNN for Affective Touch Classification in Soft Plush Companions

公开信息显示，该条目聚焦“Design and Validation of a Lightweight 1D CNN for Affective Touch Classification in Soft Plush Companions”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.AIDesign and Validation of a Lightweight 1D CNN for Affective Touch Classification in Soft Plush Companions
arXiv cs.LG1 个来源

DocOCR-Eval: A Correction-Based Framework for OCR Tool Selection Without Ground Truth

公开信息显示，该条目聚焦“DocOCR-Eval: A Correction-Based Framework for OCR Tool Selection Without Ground Truth”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.LGDocOCR-Eval: A Correction-Based Framework for OCR Tool Selection Without Ground Truth
arXiv cs.LG1 个来源

Reinforcement Learning-Guided NSGA-II Enhanced with Gray Relational Coefficient for Multi-Objective Optimization: Application to NASDAQ Portfolio Optimization

公开信息显示，该条目聚焦“Reinforcement Learning-Guided NSGA-II Enhanced with Gray Relational Coefficient for Multi-Objective Optimization: Application to NASDAQ Portfolio Optimization”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.LGReinforcement Learning-Guided NSGA-II Enhanced with Gray Relational Coefficient for Multi-Objective Optimization: Application to NASDAQ Portfolio Optimization
arXiv cs.CV1 个来源

Rethinking the Readout: Unlocking Video Backbones for AI-Generated Video Detection

公开信息显示，该条目聚焦“Rethinking the Readout: Unlocking Video Backbones for AI-Generated Video Detection”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CVRethinking the Readout: Unlocking Video Backbones for AI-Generated Video Detection
arXiv cs.CL1 个来源

Committed Before Reasoning: Behavioral Reproduction and Preliminary refresh-Level Evidence of Answer Pre-Commitment in an Open-Weight LLM

公开信息显示，该条目聚焦“Committed Before Reasoning: Behavioral Reproduction and Preliminary refresh-Level Evidence of Answer Pre-Commitment in an Open-Weight LLM”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLCommitted Before Reasoning: Behavioral Reproduction and Preliminary refresh-Level Evidence of Answer Pre-Commitment in an Open-Weight LLM
arXiv cs.CL1 个来源

Multi-level context Modeling for consistent expert selection in Mixture-of-Experts

公开信息显示，该条目聚焦“Multi-level context Modeling for consistent expert selection in Mixture-of-Experts”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLMulti-level context Modeling for consistent expert selection in Mixture-of-Experts
arXiv cs.CL1 个来源

RIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation

公开信息显示，该条目聚焦“RIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CLRIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation
arXiv cs.CV1 个来源

Training-Free Open-Vocabulary 3D Point-Cloud Segmentation on the Generalized Few-Shot Benchmark

公开信息显示，该条目聚焦“Training-Free Open-Vocabulary 3D Point-Cloud Segmentation on the Generalized Few-Shot Benchmark”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.CVTraining-Free Open-Vocabulary 3D Point-Cloud Segmentation on the Generalized Few-Shot Benchmark
OpenAI News1 个来源

Safety and alignment in an era of long-horizon models

公开信息显示，该条目聚焦“Safety and alignment in an era of long-horizon models”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
OpenAI NewsSafety and alignment in an era of long-horizon modelsOpenAI NewsDavid Vélez and Robin Vince join the boards of the OpenAI Foundation and OpenAI Group PBC
Ollama1 个来源

v0.32.2

公开信息目前只提供“v0.32.2”的标题与页面元数据，更多细节仍需回到原始来源核对。

开源1 个来源
来源与时间线
Ollamav0.32.2

7月20日周一

arXiv cs.AI1 个来源

Causal-Audit: Explicit and Auditable Graph-based Reasoning via Target-Aware Causal Chain Construction

公开信息显示，该条目聚焦“Causal-Audit: Explicit and Auditable Graph-based Reasoning via Target-Aware Causal Chain Construction”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
arXiv cs.AICausal-Audit: Explicit and Auditable Graph-based Reasoning via Target-Aware Causal Chain Construction
arXiv cs.AI1 个来源

GraphDx: A Cost-Aware Knowledge-Enhanced Multi-Agent Framework for Sequential Diagnosis

公开信息显示，该条目聚焦“GraphDx: A Cost-Aware Knowledge-Enhanced Multi-Agent Framework for Sequential Diagnosis”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
arXiv cs.AIGraphDx: A Cost-Aware Knowledge-Enhanced Multi-Agent Framework for Sequential Diagnosis
arXiv cs.CL1 个来源

VarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs

公开信息显示，该条目聚焦“VarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

研究1 个来源
来源与时间线
arXiv cs.CLVarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs
Coatue1 个来源

Our Partnership with Databricks

Coatue 发布博客回顾其与 Databricks 的合作。自 2019 年首次投资以来，Databricks 已从 Spark 数据处理平台演变为企业将 AI 嵌入工作流的关键上下文层。文章总结了 Databricks 的发展阶段：基于 Iceberg 的开放数据基础、统一湖仓架构、以及支持企业构建智能体应用（Agentic Applications）的界面。核心原语包括 Unity Catalog（数据与 AI 资产治理）和 AI Gateway（管理 token 消耗与智能体治理）。Databricks 还像研究实验室一样加速研发，将查询引擎等项目的开发周期从 3-4 年压缩至 6-9 个月。

观点1 个来源
来源与时间线
CoatueOur Partnership with Databricks

7月18日周六

7月17日周五

VentureBeat AI1 个来源

The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs

公开信息显示，该条目聚焦“The AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基准1 个来源
来源与时间线
VentureBeat AIThe AI compute gap: Enterprises are buying infrastructure faster than they can measure what it costs
VentureBeat AI1 个来源

The agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials

公开信息显示，该条目聚焦“The agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

安全1 个来源
来源与时间线
VentureBeat AIThe agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials
VentureBeat AI1 个来源

The AI context gap: Enterprise AI organizations have a trust problem, not a retrieval problem — and most are still building the fix

公开信息显示，该条目聚焦“The AI context gap: Enterprise AI organizations have a trust problem, not a retrieval problem — and most are still building the fix”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
VentureBeat AIThe AI context gap: Enterprise AI organizations have a trust problem, not a retrieval problem — and most are still building the fix
Microsoft Foundry Blog1 个来源

Answers You Can Trust: Grounding Enterprise Agents with Foundry IQ | Microsoft Community Hub

公开信息显示，该条目聚焦“Answers You Can Trust: Grounding Enterprise Agents with Foundry IQ | Microsoft Community Hub”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Microsoft Foundry BlogAnswers You Can Trust: Grounding Enterprise Agents with Foundry IQ | Microsoft Community HubMicrosoft Foundry BlogAgents can learn with Memory in Microsoft Foundry Agent Service | Microsoft Community Hub

7月16日周四

Hugging Face Transformers1 个来源

Patch release: v5.14.1

Hugging Face Transformers v5.14.1 是一个补丁版本，修复了 Inkling 模型集成中的几个问题，包括使用 EncoderDecoderCache 的辅助解码（assisted decoding）以及使用 position_bias 的 StaticCache/sdpa 预填充。还包含 FP8 内核版本更新和 deepgemm 多设备修复。

监管1 个来源
来源与时间线
Hugging Face TransformersPatch release: v5.14.1
Ollama1 个来源

v0.32.1

公开信息显示，该条目聚焦“v0.32.1”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Ollamav0.32.1
NVIDIA Generative AI Technical Blog1 个来源

Build a Multi-Camera 3D Tracking Application with NVIDIA DeepStream 9.1 Skills

NVIDIA发布了关于使用DeepStream 9.1构建多摄像头3D跟踪应用的教程。该应用旨在解决大型空间中跨摄像头跟踪同一物体的问题，超越了单摄像头2D跟踪的限制。

产品更新1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogBuild a Multi-Camera 3D Tracking Application with NVIDIA DeepStream 9.1 Skills
VentureBeat AI1 个来源

Agentic orchestration: Enterprise AI organizations have a deployment problem, not a platform problem — and most are calling chatbots agents

VentureBeat Pulse Research对101家企业调查发现，企业AI代理编排正快速向模型提供商平台集中，Anthropic的Claude以40%的使用率领先，但其正面临部署现实与雄心的巨大差距：71%的企业表示其部署的“代理”中只有不到四分之一是真正的多步骤编排工作流，多数仍为聊天机器人封装。企业因规避供应商锁定而倾向于混合控制平面（51%），仅6%接受纯厂商管理，同时实时成本控制普遍缺失（27%无应对机制）。VentureBeat指出，该样本为2026年6月单次自选调查，结果仅具方向性参考价值。

智能体1 个来源
来源与时间线
VentureBeat AIAgentic orchestration: Enterprise AI organizations have a deployment problem, not a platform problem — and most are calling chatbots agents
The Verge AI1 个来源

xAI sues a man for using Grok to generate CSAM ‘deepfakes’

马斯克旗下xAI公司起诉一名南卡罗来纳州男子，指控其使用Grok AI聊天机器人生成和传播儿童性虐待材料（CSAM）。诉状称该男子故意绕过安全措施、篡改非自愿图像并生成CSAM。案件由The Verge于2026年7月15日报道，引用路透社消息。

监管1 个来源
来源与时间线
The Verge AIxAI sues a man for using Grok to generate CSAM ‘deepfakes’
TechCrunch AI · Ars Technica AI2 个来源

Amid hardware legal battle, OpenAI releases a $230 keyboard for Codex

在与苹果的硬件商业机密窃取法律纠纷中，OpenAI发布了一款售价230美元的发光键盘，专为其代理式编程应用Codex设计。

产品更新2 个来源
来源与时间线
TechCrunch AIAmid hardware legal battle, OpenAI releases a $230 keyboard for CodexArs Technica AIOpenAI's first branded hardware is... a light-up keyboard?
Hugging Face Transformers1 个来源

Release v5.14.0

Hugging Face Transformers 发布 v5.14.0 版本，新增两个模型：Inkling 和 TIPSv2。Inkling 来自 Thinking Machines，总参数量 975B，激活参数 41B，是一个多模态模型，可接受文本、图像和音频输入并生成文本输出，适用于多种自然语言和编程语言，支持代理/工具使用系统、编程助手、聊天机器人、检索增强生成等场景，采用开放权重。TIPSv2 的描述在原始文本中不完整，仅提及名称和一个截断的文档链接。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease v5.14.0
OpenAI News · MIT Technology Review AI2 个来源

GPT-Red: Unlocking Self-Improvement for Robustness

OpenAI发布了GPT-Red，一种利用自我对弈自动进行红队测试的系统，旨在提升AI的安全性、对齐性和提示注入鲁棒性。

安全2 个来源
来源与时间线
OpenAI NewsGPT-Red: Unlocking Self-Improvement for RobustnessMIT Technology Review AIMeet GPT-Red: an LLM super-hacker OpenAI built to make its models safer

7月15日周三

arXiv cs.LG1 个来源

OmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes

OmniPM-Net是一种基于卷积条件神经过程（ConvCNP）的融合模型，旨在协调离散站点和网格化的PM10预测。它通过地形感知的高斯集合卷积将图神经网络（GNN）的站点预报提升到规则网格，再与CAMS预报融合，最终输出一致性的站点或网格预测。在中国1618个站点2024年全年的评估中，OmniPM-Net在站点精度上与更强的GNN基线相当（MAE 21.14 vs 22.00 µg/m³），同时将CAMS的MAE降低30%，并生成离散GNN无法提供的网格化场。在高浓度尾部（90百分位MAE相比GNN降低9%，相比CAMS降低25%）和沙尘事件期间表现尤为突出。

研究1 个来源
来源与时间线
arXiv cs.LGOmniPMNet: Bridging discrete and gridded PM10 forecasts via omni-query neural processes
arXiv cs.CV1 个来源

Anomalous Frame Detection Using VLM-Based Description Comparison for Extracting Expert-Specific Actions and Contextual Decision-Making Scenes with Intra-Video Self-Similarity

该论文提出一种基于视觉语言模型（VLM）的异常帧检测方法，用于从任务视频中自动提取专家特定动作和情境决策场景。方法通过帧间描述比较和视频内自相似性分析，在27个模拟配电盘维护场景中，动作候选提取率达65%，决策场景候选提取率达61%，优于传统方法的59%和33%。

研究1 个来源
来源与时间线
arXiv cs.CVAnomalous Frame Detection Using VLM-Based Description Comparison for Extracting Expert-Specific Actions and Contextual Decision-Making Scenes with Intra-Video Self-Similarity
arXiv cs.CL1 个来源

G-SHARE: A Guideline-Based Structured Reasoning Framework for Human-Factor Event Diagnosis

G-SHARE是一个基于指南的结构化推理框架，用于核电厂人因事件诊断。它将CNNP九步人因事件诊断指南转化为证据提取、逐步诊断推理和一致性修复的多阶段流程。在真实报告数据集上的评估表明，其性能优于一次性提示和传统机器学习基线，达到了更高的准确率和宏F1分数。结构化推理和一致性约束对于鲁棒诊断至关重要。

研究1 个来源
来源与时间线
arXiv cs.CLG-SHARE: A Guideline-Based Structured Reasoning Framework for Human-Factor Event Diagnosis
arXiv cs.CV1 个来源

TSCA-Net: Temporal-Spatial Clique Attention for Interpretable Multimodal Pedestrian Trajectory Prediction

TSCA-Net提出了一种用于多模态行人轨迹预测的时空团注意网络，包含三个模块：TSCA（可学习时间门控的团基目标-历史交互）、CPCP（动态团势框架下的非对称成对关系建模）和AKGR（基于目标分布熵自适应调整KAN-LSTM解码器B样条网格分辨率）。在ETH/UCY和SDD数据集上达到SOTA性能（ADE/FDE分别为0.13/0.20米和6.95/10.43像素）。

研究1 个来源
来源与时间线
arXiv cs.CVTSCA-Net: Temporal-Spatial Clique Attention for Interpretable Multimodal Pedestrian Trajectory Prediction
arXiv cs.CL1 个来源

CANDI: Contextual Alignment for Niche Domains Question Answering

论文介绍了CANDI-QA数据集，用于评估大语言模型在专业领域（如医疗、金融）中的上下文敏感问答能力。数据集包含专家整理的两种问题：信息辅助型（事实提取）和应用推理型（多跳推理）。评估了十多种语言模型，并提出了神经符号基线MTSS-Net。研究发现当前LLMs在缺乏上下文或符号集成时难以实现上下文对齐。

基准1 个来源
来源与时间线
arXiv cs.CLCANDI: Contextual Alignment for Niche Domains Question Answering
arXiv cs.CV1 个来源

GenDiff: A Dose and Anatomy Aware Diffusion Model with Structural Prior Refinement for Low-Dose CT Reconstruction and Generalization

GenDiff是一种可泛化的扩散框架，用于低剂量CT重建，能够联合建模连续辐射剂量和解剖信息。它集成了剂量-解剖编码器、剂量和解剖条件冷扩散主干、物理一致性更新以及结构先验细化模块（SPRM）。在多解剖临床数据集上的实验，包括未见过的超低剂量条件和分布外数据集，表明其优于最先进的方法。

研究1 个来源
来源与时间线
arXiv cs.CVGenDiff: A Dose and Anatomy Aware Diffusion Model with Structural Prior Refinement for Low-Dose CT Reconstruction and Generalization
arXiv cs.LG1 个来源

Semidirect Fourier Delta Attention: Phase-Controlled Delta Memory with Constructive Chunk-WY Kernels

本文提出相位控制傅里叶Delta注意力（SFDA），是Kimi Delta注意力的推广，用块旋转傅里叶控制替代实对角衰减。主要理论贡献是构造性chunk-WY因式分解，实现精确仿射块传递、稳定性与复杂度界，以及相位加低秩内存的紧凑刻画。实验表明SFDA能在玩具状态跟踪中学习循环记忆，而相位禁用的基线KDA接近随机。大型语言模型比较留待未来。

研究1 个来源
来源与时间线
arXiv cs.LGSemidirect Fourier Delta Attention: Phase-Controlled Delta Memory with Constructive Chunk-WY Kernels
arXiv cs.LG1 个来源

Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry

本文针对单次条纹投影轮廓测量（FPP）网络中的形状先验捷径问题，提出PhiCalNet。该网络输出包裹相位表示并通过固定可微标定层映射到深度，从而在架构上消除捷径。在合成基准测试中，PhiCalNet将物体平均绝对误差从14.54 mm降至4.46 mm（提升3.3倍），并首次为FPP引入逐像素共形不确定性量化。

研究1 个来源
来源与时间线
arXiv cs.LGRepairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry
arXiv cs.CL1 个来源

Scaling Point-in-Time Language Models

本文提出通过大规模训练缩小点时间语言模型与无限制模型之间的性能差距。研究者构建了参数量达4B、在1万亿时间过滤令牌上训练的decoder-only transformer，生成了2013-2024年的月度模型检查点。在常识推理和语言理解基准上，模型性能接近Gemma-3-4B和LLaMA-7B等同类模型，但仍有差距。指令微调进一步提升了可用性，并发布了完整管线。

开源1 个来源
来源与时间线
arXiv cs.CLScaling Point-in-Time Language Models
OGX1 个来源

v1.2.1

公开信息显示，该条目聚焦“v1.2.1”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OGXv1.2.1
NVIDIA Generative AI Technical Blog1 个来源

Lessons From the Leaderboard: What 5,000+ Kagglers Taught Us About Improving AI Reasoning

这篇文章总结了NVIDIA Nemotron模型推理挑战赛的经验，超过5000名Kaggle参赛者探索了提高AI推理准确性的技术。由于文本截断，无法获取完整内容。

基准1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogLessons From the Leaderboard: What 5,000+ Kagglers Taught Us About Improving AI Reasoning
NVIDIA Generative AI Technical Blog1 个来源

How to Run an Autoresearch Workflow with RL Agent Skills and NVIDIA NeMo

NVIDIA博客介绍了如何利用RL智能体技能和NVIDIA NeMo运行自动研究工作流，强调编码AI代理现在能够通过检查仓库、设置运行环境和解决问题来管理长时间运行的机器学习工作流。

监管1 个来源
来源与时间线
NVIDIA Generative AI Technical BlogHow to Run an Autoresearch Workflow with RL Agent Skills and NVIDIA NeMo

7月14日周二

vLLM1 个来源

v0.25.1

vLLM v0.25.1 补丁版本发布，包含两个针对性 bug 修复：避免缺少系统 FFmpeg 时 TorchCodec 导入错误阻塞模型启动（例如 vllm serve Qwen/Qwen3-VL-2B-Instruct），并为混合精度的 allreduce + RMSNorm + 静态量化融合增加类型匹配保护，防止隐藏状态损坏。

基础设施1 个来源
来源与时间线
vLLMv0.25.1
SGLang1 个来源

v0.5.15.post1

SGLang v0.5.15.post1 发布，包含若干补丁，主要修复 GLM 5.2 的 IndexShare 问题（PD 分离和上下文并行设置），以及 DSA 模型在非 CUDA/HIP 设备上的启动、CUDA 12 镜像的 flashinfer 依赖、flashinfer trtllm FP4 MoE 内核在长输入下产生 NaN 输出等问题。

基础设施1 个来源
来源与时间线
SGLangv0.5.15.post1
llama.cpp1 个来源

b9994

llama.cpp 发布 b9994 版本，主要新增 Metal 后端对 Q2_0 量化格式的支持（PR #25419）。此外，macOS Apple Silicon 的 KleidiAI 版本已被禁用（PR #23780）。

基础设施1 个来源
来源与时间线
llama.cppb9994
arXiv cs.CL1 个来源

Index SLM Technical Report

哔哩哔哩发布了Index-1.9B系列开源小语言模型，包括Base（19亿参数，预训练于2.8万亿token）、Pure、Chat和Character。Base模型在基准测试中平均得分64.92，与更大模型竞争。预训练采用Warmup-Stable-Decay学习率计划和Norm-Head输出层。对照研究揭示了模型深度、学习率、数据质量的影响，以及一个无法解释的基准性能激增。模型和代码已开源。

基准1 个来源
来源与时间线
arXiv cs.CLIndex SLM Technical Report
arXiv cs.LG1 个来源

AuditWeave: A Tamper-Evident, Auditor-Navigable Evidence Layer for AI-Assisted and Data-Transformation Workflows

AuditWeave是一个轻量级Python库，通过追加式哈希链账本记录AI辅助和数据转换工作流，支持审计追踪，检测篡改，性能开销为每事件数十微秒。

研究1 个来源
来源与时间线
arXiv cs.LGAuditWeave: A Tamper-Evident, Auditor-Navigable Evidence Layer for AI-Assisted and Data-Transformation Workflows
arXiv cs.CL1 个来源

CLIR-Bench: Benchmarking Multimodal Question Answering over Irregular Clinical Time Series

CLIR-Bench是一个针对不规则临床时间序列的多模态问答基准，由去标识化的ICU记录构建，包含6,600个实例，覆盖11个临床变量，组织成4个能力维度和11个任务。实验表明，现有通用模型在检索和推理稀疏临床证据方面存在困难，凸显了更强的不规则时间序列推理方法的必要性。

基准1 个来源
来源与时间线
arXiv cs.CLCLIR-Bench: Benchmarking Multimodal Question Answering over Irregular Clinical Time Series
arXiv cs.AI1 个来源

Faithful, Not Corrective: Message-Format Effects in Multi-Hop Agent Relays Are Tier-Dependent

该论文研究了多跳LLM代理中消息格式（自由自然语言、精确指导自然语言、JSON、三元组、键值对）对信息传递保真度的影响。通过程序化生成的原子事实在六跳中重新编码的受控中继测试平台，发现格式效应依赖于中继能力：强中继在忠实指令下几乎无损，格式和认知负荷影响极小；弱中继（1.5B参数）下，格式间六跳召回率差异扩大8.7倍，其中JSON的固定键模式提供漂移阻力但需支付编码代价；错误注入后，错误值在83-100%的链中持续存在至最终跳，且无附带损害。结构提供的是忠实且错误定位的通道，而非纠错码，格式选择应遵循管道中最弱的中继。

智能体1 个来源
来源与时间线
arXiv cs.AIFaithful, Not Corrective: Message-Format Effects in Multi-Hop Agent Relays Are Tier-Dependent
arXiv cs.AI1 个来源

Format Sensitivity Index: Token-Controlled Prompt Wrapper Robustness and Schema Compliance in LLM Benchmarking

本文提出格式敏感度指数（FSI）和可解析性敏感度指数（PSI），用于量化提示包装器格式变化对LLM性能和输出合规性的影响。在140,000次OpenRouter生成实验中发现，不同模型的FSI差异超过30倍，且可解析性是准确率的重要预测因子。作者认为仅报告准确率而不考虑包装器方差和合规性是不稳健的，并给出了基准测试和结构化输出部署的建议。

研究1 个来源
来源与时间线
arXiv cs.AIFormat Sensitivity Index: Token-Controlled Prompt Wrapper Robustness and Schema Compliance in LLM Benchmarking
arXiv cs.AI1 个来源

From ML Predictions to Informed Diagnostic Assistance Using the Toulmin Model of Argumentation

本文提出一种基于图尔敏论证模型的医学影像诊断框架，将机器学习（ML）对视网膜诊断的预测分解为主张、依据、保证、限定词、反驳和支持等组件。通过专用模型提取生物标志物作为依据，MedGemma代理分析保证，基于定量评估确定限定词，并利用MedSigLip计算图像相似度构建反驳，最终向人类专家呈现结构化、可解释的诊断辅助信息。

智能体1 个来源
来源与时间线
arXiv cs.AIFrom ML Predictions to Informed Diagnostic Assistance Using the Toulmin Model of Argumentation
arXiv cs.LG1 个来源

Knowledge Graphs Meet Graph Neural Networks: A Comprehensive Survey

该论文提出了一种面向知识图谱中图神经网络（GNN）应用的双层分类体系，涵盖知识图谱构建、嵌入、推理及应用全流程，并按照GCN、GAT、HGNN等模型类型进行划分。综述了多种基于GNN的模型，分析优势与局限，并讨论了未解决的挑战与未来研究方向。

研究1 个来源
来源与时间线
arXiv cs.LGKnowledge Graphs Meet Graph Neural Networks: A Comprehensive Survey
arXiv cs.LG1 个来源

Position: Every Ground Truth is a Human Construction, not an Objective Truth

本文是arXiv上的一篇立场论文，指出机器学习中的ground truth数据集并非客观中立，而是由人类和技术共同构建的。作者主张应明确讨论这些数据集的情境依赖性和局限性，以提高模型的可靠性、透明度和问责制。

研究1 个来源
来源与时间线
arXiv cs.LGPosition: Every Ground Truth is a Human Construction, not an Objective Truth
arXiv cs.CV1 个来源

Cross-Subject Modeling for Widefield Calcium Imaging via Atlas-Aligned Spatiotemporal Tokenization

该论文提出WiCAT，一种用于宽场钙成像数据的多主体模型，通过自监督预训练和基于图谱的令牌化方案，实现跨主体、跨任务和跨数据集的泛化，并在未见主体上实现零样本连续行为解码和脑区重建。

研究1 个来源
来源与时间线
arXiv cs.CVCross-Subject Modeling for Widefield Calcium Imaging via Atlas-Aligned Spatiotemporal Tokenization
arXiv cs.CL1 个来源

RouteRec: Strict Evaluation of Recommender-Agent Selection and Aggregation

论文《RouteRec: 推荐系统智能体选择与聚合的严格评估》于2026年7月14日发表在arXiv cs.CL上。该研究提出RouteRec框架，对比了请求级硬选择与物品级学习聚合两种策略在四种传统推荐智能体和一种基于LLM的重排序器上的表现。在MovieLens-1M数据集上，使用无泄漏的5折外协议时，硬选择效果不及BM25（HR@10=0.223 vs 0.254），而学习聚合的廉价版本在HR指标上与BM25持平，NDCG略高（0.123 vs 0.114），门控全智能体聚合则取得了HR@10=0.295，但需要调用70.2%的LLM。结论是请求级选择过于粗糙，物品级聚合更值得探索。

研究1 个来源
来源与时间线
arXiv cs.CLRouteRec: Strict Evaluation of Recommender-Agent Selection and Aggregation
TechCrunch AI1 个来源

Uber’s product chief on hotels, robotaxis, and why the company doesn’t want to be “everything for everyone”

Uber首席产品官Sachin Kansal接受TechCrunch采访，讨论了公司在金融服务、与Waymo日益复杂的关系、新的AV Labs数据运营，以及AI如何实际影响乘客和司机体验等方面的计划。标题还涉及酒店和机器人出租车业务，但文章强调Uber不愿成为“万能平台”。

媒体/访谈1 个来源
来源与时间线
TechCrunch AIUber’s product chief on hotels, robotaxis, and why the company doesn’t want to be “everything for everyone”
llama.cpp1 个来源

b9993

llama.cpp 发布 b9993 版本，主要新增对腾讯 Hunyuan 3（Hy3/hy_v3）模型架构的支持，并集成了 MTP 推测解码。该模型为 MoE 解码器堆栈，包含每头 Q/K RMSNorm、sigmoid 路由器、始终激活的未门控共享专家以及前导密集块。实现基于 charlie12345 的分支，为兼容性将 blk.N.exp_probs_b 存储为无 .bias 后缀。同时提供 macOS Apple Silicon 和 Intel 的二进制下载。

基础设施1 个来源
来源与时间线
llama.cppb9993
llama.cpp1 个来源

b9992

llama.cpp 发布 b9992 版本，主要重构了 CUDA MMQ 内核配置（#24127），包括修复 Blackwell 配置和移除遗留代码。此外，macOS Apple Silicon (arm64) 的 KleidiAI 启用版本已被禁用。提供多平台二进制下载。

基础设施1 个来源
来源与时间线
llama.cppb9992
The Verge AI · Ars Technica AI2 个来源

The 6 wildest claims in Apple’s lawsuit against OpenAI

苹果起诉OpenAI，指控其在招聘面试中要求苹果员工携带未发布的硬件组件和样品，并窃取机密文件、监视硬件原型。

监管2 个来源
来源与时间线
The Verge AIThe 6 wildest claims in Apple’s lawsuit against OpenAIArs Technica AIApple sues OpenAI after ex-engineer allegedly used bug to steal trade secrets

7月13日周一

7月12日周日

vLLM1 个来源

v0.25.0

公开信息显示，该条目聚焦“v0.25.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
vLLMv0.25.0

7月11日周六

Ollama1 个来源

v0.32.0

公开信息显示，该条目聚焦“v0.32.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Ollamav0.32.0
SGLang1 个来源

v0.5.15

公开信息显示，该条目聚焦“v0.5.15”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
SGLangv0.5.15

7月10日周五

OGX1 个来源

v1.2.0

公开信息显示，该条目聚焦“v1.2.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
OGXv1.2.0
MIT Technology Review AI · Anthropic Research2 个来源

Anthropic found a hidden space where Claude puzzles over concepts

Anthropic 开发了一种名为 Jacobian 透镜的技术，首次清晰揭示了大型语言模型（如 Claude）在回答问题或执行任务时的内部运作机制，结果从平凡到令人不安。

研究2 个来源
来源与时间线
MIT Technology Review AIAnthropic found a hidden space where Claude puzzles over conceptsAnthropic ResearchA global workspace in language models
OpenAI Python SDK1 个来源

v2.45.0

公开信息显示，该条目聚焦“v2.45.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.45.0
Anthropic Research1 个来源

How Claude Performs on Robotics Tasks

Anthropic 研究团队测试了语言模型（如 Claude）在机器人任务中的表现。他们让模型通过不同接口控制多种机器人（如四足机器人、机械臂等），包括直接控制关节力矩、编写控制器代码、使用预训练策略以及强化学习训练。结果发现，模型在直接低层控制上表现不佳，但在使用预训练策略或高层指令时能完成导航和操作任务。新一代模型能力提升显著，但尚无法在没有预训练策略的情况下控制人形机器人。

研究1 个来源
来源与时间线
Anthropic ResearchHow Claude Performs on Robotics Tasks

7月9日周四

Anthropic Research1 个来源

An off switch for dual use knowledge in AI models

Anthropic与AE Studio合作提出GRAM（梯度路由辅助模块）方法，为AI模型添加可移除的组件以管理双重用途知识（如病毒学、网络安全）。该方法允许在单个模型中精确控制危险能力，无需训练多个独立模型。研究尚处于初步阶段，未应用于Anthropic的生产模型。

安全1 个来源
来源与时间线
Anthropic ResearchAn off switch for dual use knowledge in AI models
PyTorch1 个来源

PyTorch 2.13.0 Release

PyTorch 2.13.0 发布，要点包括：FlexAttention 登陆 Apple Silicon (MPS)，在稀疏模式上相对 SDPA 最高可提速约 12 倍，并在 CUDA 上新增确定性反向路径；CuTeDSL“Native DSL”后端为 Inductor 提供继 Triton 之后的第二条高性能 GPU 代码路径（原型），编译更快；nn.LinearCrossEntropyLoss 将最终预测与损失计算合并，可在大规模词表语言模型训练中使峰值 GPU 内存最高减少 4 倍；torchcomms 作为 PyTorch Distributed 的新通信后端，提升了大规模集群训练的容错性、可扩展性和可调试性。

开源1 个来源
来源与时间线
PyTorchPyTorch 2.13.0 Release

7月8日周三

NVIDIA AI Blog1 个来源

NVIDIA Nemotron Achieves Benchmark-Leading Performance With LangChain Deep Agents Harness

NVIDIA Nemotron 3 Ultra 与 LangChain 的 Deep Agents 框架集成后，在基准测试中表现出领先水平，且成本低于顶级闭源模型。该组合在开源模型中达到最高准确率，并能以更高吞吐量完成更多任务；原文提到运行速度为 10 倍（内容截断，具体指代不明）。

智能体1 个来源
来源与时间线
NVIDIA AI BlogNVIDIA Nemotron Achieves Benchmark-Leading Performance With LangChain Deep Agents Harness

7月7日周二

MIT Technology Review AI1 个来源

The foundational elements of AI architecture that IT leaders need to scale

文章探讨了AI架构的基础要素，帮助IT领导者在快速发展的AI能力（包括向智能体系统迁移）中做出可扩展的投资决策，同时管理风险。原文截断，仅提供开头部分。

观点1 个来源
来源与时间线
MIT Technology Review AIThe foundational elements of AI architecture that IT leaders need to scale
Microsoft Semantic Kernel1 个来源

python-1.44.0

微软语义内核（Microsoft Semantic Kernel）发布了 Python 1.44.0 版本。此版本主要包含依赖升级：Python 依赖（tornado、pyjwt、starlette、bleach、pyarrow）以及 .NET 依赖（Aspire.Azure.Search.Documents、Aspire.Hosting.Azure.CognitiveServices 和前端相关依赖）。

开源1 个来源
来源与时间线
Microsoft Semantic Kernelpython-1.44.0
Ollama1 个来源

v0.31.2

Ollama v0.31.2 发布，主要更新包括：在旧款NVIDIA GPU（计算能力6.x）上启用flash attention；集成显卡现在可以通过填充来适配显存以卸载视觉模型；修复思维模型在禁用思维时的结构化输出问题；强化GGUF模型创建；`ollama launch`为Claude Code默认禁用遥测；修复非UTF-8路径上的模型加载问题；更新MLX和llama.cpp引擎。新贡献者 @kevinpark1217。

开源1 个来源
来源与时间线
Ollamav0.31.2

7月4日周六

Hugging Face Transformers1 个来源

Release v5.13.0

Hugging Face Transformers 发布 v5.13.0，新增 KimiK 2.5、2.6 和 2.7 模型架构，基于 KimiK 2.5 的开源原生多模态智能体模型，专注于长代码任务、代码驱动设计、自主执行及集群编排，支持 Rust、Go、Python 等多种编程语言和全栈开发。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease v5.13.0

7月3日周五

Anthropic Python SDK1 个来源

v0.116.0

公开信息显示，该条目聚焦“v0.116.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

智能体1 个来源
来源与时间线
Anthropic Python SDKv0.116.0

7月2日周四

7月1日周三

Ollama1 个来源

v0.31.1

Ollama v0.31.1 发布，主要改进了 Apple Silicon 上 Gemma 4 的性能，通过多令牌预测（MTP）平均加速近 90%，无需配置即可自动优化。其他更新包括收紧 MLX 引擎中的 Gemma 4 MoE 模型加载、更新 MLX 引擎和 llama.cpp 引擎。

基础设施1 个来源
来源与时间线
Ollamav0.31.1
Anthropic Python SDK1 个来源

v0.115.0

Anthropic Python SDK v0.115.0 发布，新增对托管代理事件增量流、代理覆盖、反向分页、保险库凭证注入作用域以及代理和部署webhook事件的支持。

开源1 个来源
来源与时间线
Anthropic Python SDKv0.115.0
Anthropic News1 个来源

Introducing Claude Sonnet 5

Anthropic推出Claude Sonnet 5，称为迄今最具代理性的Sonnet模型，在编程和日常专业工作中具有顶级智能。其性能接近Opus 4.8，但价格更低，相较Sonnet 4.6在推理、工具使用、编程和知识工作方面有显著提升。该模型即日起在所有计划中可用，定价为每百万输入tokens 2美元、每百万输出tokens 10美元。安全评估显示其不良行为率低于Sonnet 4.6，早期合作伙伴反馈其在复杂任务中的自主性更强。

安全1 个来源
来源与时间线
Anthropic NewsIntroducing Claude Sonnet 5

6月30日周二

vLLM1 个来源

v0.24.0

公开信息显示，该条目聚焦“v0.24.0”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
vLLMv0.24.0

6月27日周六

SGLang1 个来源

v0.5.14

公开信息显示，该条目聚焦“v0.5.14”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

基础设施1 个来源
来源与时间线
SGLangv0.5.14
OGX1 个来源

v0.5.4

公开信息显示，该条目聚焦“v0.5.4”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
OGXv0.5.4

6月25日周四

6月24日周三

OGX1 个来源

v1.1.3

OGX 发布 v1.1.3 版本，包含两个修复：修复 Vertex AI 中路由表遍历以重置提供者客户端（回溯 #6148），修复 pgvector 中在创建连接池前确保向量扩展存在（回溯 #6168）；以及更新 UI 锁文件中的 ogx-client 依赖。

基础设施1 个来源
来源与时间线
OGXv1.1.3

6月18日周四

PyTorch1 个来源

PyTorch 2.12.1 Release, bug fix release

PyTorch 2.12.1 是错误修复版本，修复了若干回归和静默正确性问题：通过更新 Triton 至 3.7.1 修复了 NVIDIA B200 GPU 上 Flash Attention 的非确定性输出以及 B100/B200 (sm100) GPU 上 Triton convolution2d_bwd_weight 内核的非法内存访问；修复了字节类型视图上 fill_ 操作存储偏移未对齐的问题；并从二进制构建矩阵中移除了 CPython 3.13t。

监管1 个来源
来源与时间线
PyTorchPyTorch 2.12.1 Release, bug fix release

6月17日周三

OGX1 个来源

v1.1.2

OGX 发布 v1.1.2 版本，主要更新包括：更新 UI 锁文件中的 ogx-client 至 ^1.1.1，修复对话中孤立项的级联删除，以及为 MarkItDown 处理器添加 ZIP 解压限制。

开源1 个来源
来源与时间线
OGXv1.1.2

6月16日周二

6月15日周一

OGX1 个来源

v1.1.1

OGX 发布 v1.1.1 版本，主要修复了文件处理器同步解析、Milvus 兼容性、VertexAI 运行时错误及向量搜索错误传播等问题。

开源1 个来源
来源与时间线
OGXv1.1.1
vLLM1 个来源

v0.23.0

vLLM v0.23.0 发布，包含 408 次提交和 200 位贡献者。主要亮点：DeepSeek-V4 进一步优化，包括稀疏 MLA 元数据分离、TRTLLM 内核支持等；Model Runner V2 默认用于 Llama 和 Mistral 密集模型，新增 FlashInfer 采样器等功能。注意：Minimax M3 暂不支持。

基础设施1 个来源
来源与时间线
vLLMv0.23.0

6月13日周六

SGLang1 个来源

v0.5.13

SGLang 发布 v0.5.13 版本，新增多个模型支持，包括自回归模型 Nemotron 3 Ultra、Step-3.7-Flash、Command A+，以及扩散模型 Cosmos3、LingBot-World、SANA-WM、Ernie-Image、FLUX.2-Klein 4B/9B、Ideogram 4。同时，Spec V2 成为默认的推测解码路径，树形解码（topk>1）在多种后端上达到生产就绪状态。

基础设施1 个来源
来源与时间线
SGLangv0.5.13

6月12日周五

6月11日周四

OGX1 个来源

v1.1.0

OGX 发布了 v1.1.0 版本，主要包含修复（如请求头清理、CI 测试夹具问题）、性能改进（并行化健康检查和向量存储扇出）、文档更新（1.0 发布说明、博客公告、API 端点引用修正）、依赖清理（移除未使用的 litellm）以及新增 OpenAI Responses 模式漂移检查器。

监管1 个来源
来源与时间线
OGXv1.1.0

6月10日周三

5月30日周六

OGX1 个来源

v0.4.6

OGX v0.4.6 发布，主要修复了多个安全漏洞（CVE-2026-32597、CVE-2026-30922、CVE-2026-27628、CVE-2025-14009、CVE-2026-33236、CVE-2026-48710），并更新了依赖版本（pyjwt、pyasn1、pypdf、nltk、starlette），同时添加了发布自动化工作流。

安全1 个来源
来源与时间线
OGXv0.4.6

5月28日周四

OGX1 个来源

v0.7.2

OGX 发布了 v0.7.2 版本，主要变更：更新 UI 锁文件中的 llama-stack-client 到 ^0.7.1，以及约束 starlette >=1.0.1 以修复 CVE-2026-48710 安全漏洞。

安全1 个来源
来源与时间线
OGXv0.7.2
Bessemer Venture Partners1 个来源

Researching the frontier of robotics: Three founders on what it takes to succeed in embodied AI

From the data pyramid to ‘dark magic’ in data attribution, here's what three founders at the cutting edge of embodied AI are watching closely—and what they think their field is missing.

其他1 个来源
来源与时间线
Bessemer Venture PartnersResearching the frontier of robotics: Three founders on what it takes to succeed in embodied AI

5月25日周一

arXiv cs.CV1 个来源

GEM-4D: Geometry-Enhanced Video World Models for Robot Manipulation

GEM-4D是一种几何增强的视频世界模型，通过注入从预训练几何基础模型蒸馏的密集4D对应监督，解决了视频世界模型在机器人操作中缺乏一致点级运动的问题。该模型在训练时联合学习外观和几何结构，保持单流架构且无额外推理成本。还引入逆动力学模块将对应一致的视频展开转换为可执行机器人轨迹，在真实和模拟操作中均可部署。在视频预测和几何一致性上达到最先进水平，真实世界操作成功率从61%提升至81%。

研究1 个来源
来源与时间线
arXiv cs.CVGEM-4D: Geometry-Enhanced Video World Models for Robot Manipulation
arXiv cs.AI1 个来源

BOHM: Zero-Cost Hierarchical Attribution for Compound AI Systems

本文介绍BOHM，一种零成本层次归因方法，通过解析复合AI系统已有的路由权重，构建层次归因树，无需访问组件内部或额外评估，支持多分辨率归因。实验表明，在多种任务上BOHM与Shapley方法结果高度相关，但计算成本极低。

研究1 个来源
来源与时间线
arXiv cs.AIBOHM: Zero-Cost Hierarchical Attribution for Compound AI Systems
arXiv cs.CV1 个来源

Improved Vision-to-Chart Buoy Association with Learned World-to-Image Projection

该论文提出一种轻量级改进方法，用于MaCVi 2026视觉-海图数据关联挑战中的DETR融合Transformer基线。通过训练专用MLP（QueryMLP）从海图测量和IMU数据显式预测浮标在水线处的像素坐标，并将其附加到解码器查询向量中，以提供空间先验，减轻Transformer解码器的几何推理负担。在挑战排行榜上，该方法获得总体得分0.7386，F1=0.8055，mIoU=0.6718，位列第二。

研究1 个来源
来源与时间线
arXiv cs.CVImproved Vision-to-Chart Buoy Association with Learned World-to-Image Projection
arXiv cs.AI1 个来源

NeuroNL2LTL: A Neurosymbolic Framework for Natural Language Translation of Linear Temporal Logic

NeuroNL2LTL是一个神经符号框架，用于将自然语言翻译为线性时序逻辑（LTL）。它通过中间表示实现结构保持的映射，并结合可满足性检查与最小编辑修复机制。核心创新是验证器在环训练，将验证结果作为强化学习奖励信号，直接优化形式正确性。在20万+跨领域需求上，达到28%语义等价和86%的可满足性验证率，并生成上下文解释，使领域专家无需专业培训即可验证规约。

研究1 个来源
来源与时间线
arXiv cs.AINeuroNL2LTL: A Neurosymbolic Framework for Natural Language Translation of Linear Temporal Logic
arXiv cs.LG1 个来源

Reading Calibrated Uncertainty from Language Model Trajectories

论文《从语言模型轨迹中读取校准的不确定性》提出提取十一种尺度不变的几何特征，这些特征来自逐层MLP更新的累积路径，并输入稀疏线性探针以校准不确定性。在选择性弃权设置下，该方法优于最大软最大概率（MSP），最高提升21个AURC点。几何特征可解释地追踪误差在深度网络中的形成过程。

研究1 个来源
来源与时间线
arXiv cs.LGReading Calibrated Uncertainty from Language Model Trajectories
arXiv cs.AI1 个来源

RMA: an Agentic System for Research-Level Mathematical Problems

本文提出研究数学智能体（RMA），一个用于研究级数学问题自动推理的智能体框架。RMA将证明求解分解为问题分析、文献搜索与理解、公平比较、知识库构建和证明验证等模块，由初始化、提议和验证智能体通过共享结构化内存协调。在包含10个研究级问题的First Proof基准上，RMA解决了其中8个问题，优于GPT-5.2R和Aletheia等强基线。消融实验表明性能提升源于结构化推理模块、迭代细化和验证器反馈的交互。代码将在接收后公开。

研究1 个来源
来源与时间线
arXiv cs.AIRMA: an Agentic System for Research-Level Mathematical Problems
arXiv cs.AI1 个来源

SciAtlas: A Large-Scale Knowledge Graph for Automated Scientific Research

SciAtlas是一个大规模多学科知识图谱，包含超过4300万篇论文、1.57亿实体和30亿三元组，旨在通过结构化拓扑认知基座和神经符号检索算法实现自动化科学研究，包括文献综述、趋势合成、创意定位和学术轨迹探索。

开源1 个来源
来源与时间线
arXiv cs.AISciAtlas: A Large-Scale Knowledge Graph for Automated Scientific Research
arXiv cs.LG1 个来源

The Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models

本文研究了小语言模型在算术链式思维（CoT）推理中的答案读取阶段。发现模型存在一种位置捷径：直接复制答案分隔符前的最后一个数字，而不依赖于中间推理步骤。该复制机制在1-3B参数模型的GSM8K任务中贡献了54-92个百分点的准确率（占教师强制上限的89-92%）。即使答案错误，最终答案与最后一个CoT数字匹配的概率也高达95-96%。替换末尾数字会导致准确率骤降至接近零，而移除该数字可恢复5-32个百分点。Qwen和Llama模型在87-95%的情况下会复制新引入的干扰数字，而Gemma模型会选择性门控。头部消融实验显示该行为与特定架构的头集相关。在非算术任务（BBH）上，打乱CoT步骤的保留率大幅下降；在7-8B参数模型中出现了内容选择性门控。该发现表明步骤级忠实性评估可能错误地将位置性的数字传输当作真实计算，这对基于CoT的监督机制构成挑战。

研究1 个来源
来源与时间线
arXiv cs.LGThe Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models
arXiv cs.AI1 个来源

Energy per Successful Goal: Goal-Level Energy Accounting for Agentic AI Systems

该论文提出A-LEMS框架，将AI能耗计量单位从每次推理能耗改为每个成功目标能耗（EpG），并定义了编排开销指数（OOI）。实验表明，智能体工作流每个成功目标的平均能耗是线性基线的4.33倍，且能耗主要由编排结构而非推理计算驱动。结论认为每次推理能耗不适合智能体AI系统。

智能体1 个来源
来源与时间线
arXiv cs.AIEnergy per Successful Goal: Goal-Level Energy Accounting for Agentic AI Systems
arXiv cs.LG1 个来源

FuRA: Full-Rank Parameter-Efficient Fine-Tuning with Spectral Preconditioning

该论文提出FuRA（全秩适应），一种基于谱预条件的高效全秩微调方法。通过块张量列车分解固定预训练SVD基，仅优化紧凑核心和奇异值，在保持全秩表达力的同时实现参数、内存和步时效率（与LoRA相当）。在LLM微调（LLaMA-3-8B常识推理+1.37）、数学推理强化学习和VLM视觉指令调优中超越全微调，4位量化变体QFuRA也超越QLoRA。

研究1 个来源
来源与时间线
arXiv cs.LGFuRA: Full-Rank Parameter-Efficient Fine-Tuning with Spectral Preconditioning
arXiv cs.CL1 个来源

Query-Adaptive Semantic Chunking for Retrieval-Augmented Generation: A Dynamic Strategy with Contextual Window Expansion

该论文提出了一种查询自适应语义分块（QASC）方法，用于检索增强生成（RAG）系统。QASC通过三个机制动态构建分块：查询与句子嵌入的余弦相似度评分、上下文窗口扩展以保持连贯性，以及分块级别分数聚合。在100篇技术文档和200个查询上的评估表明，QASC的F1分数达到0.85，比固定分块提高18-27%，比语义和智能体方法提高8-12%。消融研究和人工评估（Cohen kappa=0.82）验证了各组成部分的有效性。

研究1 个来源
来源与时间线
arXiv cs.CLQuery-Adaptive Semantic Chunking for Retrieval-Augmented Generation: A Dynamic Strategy with Contextual Window Expansion
arXiv cs.LG1 个来源

Latent Cache Flow: Model-to-Model Communication Without Text

该论文提出 Latent Cache Flow（LCF），一种无需文本的模型间通信方法。通过联合翻译和压缩键值缓存，将适配器大小降至先前工作 C2C 的 4%，并支持不同上下文。初步实验表明，相比基于文本的通信，LCF 准确率提高 23%，速度提升 8.5 倍。

研究1 个来源
来源与时间线
arXiv cs.LGLatent Cache Flow: Model-to-Model Communication Without Text
arXiv cs.CL1 个来源

Evaluating Large Language Models in a Complex Hidden Role Game

该论文在社交推理游戏“秘密希特勒”中评估了大型语言模型的推理、说服和欺骗能力。作者提出了开源框架和新颖指标，发现模型在复杂多轮操纵中表现不佳：规则算法与专家决策一致性达86.7%，而Llama 3.1 70B仅59.7%；链式思维提示和内部记忆未能提升表现，甚至使法西斯角色的胜率下降23.2%。研究强调了检测模型何时掌握欺骗行为的重要性，并为未来对齐研究提供了可复现测试平台。

基准1 个来源
来源与时间线
arXiv cs.CLEvaluating Large Language Models in a Complex Hidden Role Game
arXiv cs.LG1 个来源

FusionSense: Tri-Stage Near-Sensor Learning for Runtime-Adaptive Multimodal Edge Intelligence

本文提出FusionSense，一种面向能量受限自主边缘系统的融合感知智能框架。通过三阶段近传感器学习（服务器端融合模型学习任务、滤除安全标签量化模态必要性、压缩边缘融合模型），实现运行时自适应多模态决策，显著降低计算与通信开销。在RGB+深度/激光雷达双模态实验中，相比单模态过滤器，FusionSense在1%兴趣点出现率下实现33倍能耗降低，10%时11倍，固定30%数据缩减时质量损失减少92.3%，能效比最佳基线高约1.5倍。

安全1 个来源
来源与时间线
arXiv cs.LGFusionSense: Tri-Stage Near-Sensor Learning for Runtime-Adaptive Multimodal Edge Intelligence
arXiv cs.CL1 个来源

Knowledge Distillation for Low-Resource Open-source Text-to-SQL Model

该论文提出了一个知识感知的Text-to-SQL框架，通过构建包含模式语义、缩写、业务逻辑和查询模式的任务特定知识库，并将其注入到训练和推理中，从而在低资源领域特定场景下提升开放和封闭源代码大语言模型的性能。实验在七个基准测试上验证了该方法的有效性。

研究1 个来源
来源与时间线
arXiv cs.CLKnowledge Distillation for Low-Resource Open-source Text-to-SQL Model
arXiv cs.CV1 个来源

Suicide Risk Assessment from AI-powered Video Surveillance: An Interpretable Framework for Prevention in Metro Stations

该论文首次系统化定义了地铁站自杀风险评估（SRA）任务，并提出了一种可解释的AI框架。该框架结合行人追踪、活动识别、站台语义分割和轨迹驱动的风险热图建模，从监控视频中累积证据评估自杀风险，在真实监控数据上达到83.2% ROC-AUC。

研究1 个来源
来源与时间线
arXiv cs.CVSuicide Risk Assessment from AI-powered Video Surveillance: An Interpretable Framework for Prevention in Metro Stations
arXiv cs.CV1 个来源

VideoOdyssey: A Benchmark for Ultra-Long-Context and Omni-Modal Video Understanding

VideoOdyssey是一个用于超长上下文和全模态视频理解的基准测试，强调连续证书长度（必须连续观看的视频时长）。其特点包括：平均109分钟的视频时长（覆盖11个领域、54个子类别）、两个子集（VideoOdyssey-V和-AV），以及从秒到小时的5个粒度级别。评估发现当前多模态大语言模型在连续推理、细粒度感知和非语言全模态理解方面存在瓶颈。

基准1 个来源
来源与时间线
arXiv cs.CVVideoOdyssey: A Benchmark for Ultra-Long-Context and Omni-Modal Video Understanding
arXiv cs.CV1 个来源

Seeing without Looking: Do Vision-Language Benchmarks Really Test Vision?

该论文发现，在视觉语言模型（VLM）的评估基准中，即使移除大量图像标记，模型在幻觉基准上的性能下降也非常微小，表明当前基准无法可靠地测试模型对细粒度视觉证据的依赖。通过全局退化、局部遮挡、问题改写等多种实验，以及表示层分析揭示深层视觉标记相似性增加，论文认为现有基准不足以评估VLM的精细视觉接地能力。

基准1 个来源
来源与时间线
arXiv cs.CVSeeing without Looking: Do Vision-Language Benchmarks Really Test Vision?
arXiv cs.CL1 个来源

How Far Will They Go? Red-Teaming Online Influence with Large Language Models

该论文提出一个红队测试框架，用于测量开源大语言模型（LLM）在政治话题上的“奥弗顿窗口”（即模型能可靠表达的政治观点范围），并评估简单自然语言越狱如何扩大该范围。研究评估了30多个LLM，发现模型在政治表达上存在系统性不对称：开源LLM更愿意生成左倾社交媒体内容，窗口大小与模型规模负相关，且地区差异显著。越狱效果在不同模型家族间差异明显。该工作为审计开源LLM的政治可操控性提供了实用框架。

开源1 个来源
来源与时间线
arXiv cs.CLHow Far Will They Go? Red-Teaming Online Influence with Large Language Models
arXiv cs.CL1 个来源

A Survey of Text and Speech Resources for Hausa and Fongbe: Availability, Quality, and Gaps for NLP Development

该调查论文系统整理了豪萨语（约8000万-1亿 speakers）和丰贝语（约200万 speakers）的公开文本与语音资源，分析其可用性、质量和差距。豪萨语在新闻、百科和教育等领域拥有更丰富的文本资源，而丰贝语近期有学术语音数据采集项目。两种语言均包含在Masakhane基准测试中。论文指出了优先缺口，包括丰贝语的多样化文本和豪萨语的专用语音语料库。

研究1 个来源
来源与时间线
arXiv cs.CLA Survey of Text and Speech Resources for Hausa and Fongbe: Availability, Quality, and Gaps for NLP Development

5月24日周日

llama.cpp1 个来源

b9305

llama.cpp 发布 b9305 版本，通过 cmake 修复 UI 构建（添加 -fPIC 并重命名帮助程序）。提供 macOS（Apple Silicon、Intel）、iOS 及 Linux（多种架构，含 Vulkan）的二进制文件。

开源1 个来源
来源与时间线
llama.cppb9305
llama.cpp1 个来源

b9297

llama.cpp 发布 b9297 版本，主要新增对 NVFP4 MTP 缩放张量的支持，用于优化 Qwen3.5 等模型的 MTP 张量处理。同时提供了 macOS、iOS、Linux 等多个平台的预编译二进制文件。

开源1 个来源
来源与时间线
llama.cppb9297

5月23日周六

llama.cpp1 个来源

b9296

llama.cpp 发布 b9296 版本，主要修复了 ggml 库中一个关于回退 2d get 方法前检查正确 iface 方法的 bug（PR #23514）。该版本提供了 macOS（Apple Silicon 和 Intel）、iOS、Linux（Ubuntu x64/arm64/s390x，CPU 及 Vulkan）的二进制下载。

开源1 个来源
来源与时间线
llama.cppb9296

5月22日周五

arXiv cs.LG1 个来源

GraphDiffMed: Knowledge-Constrained Differential Attention with Pharmacological Graph Priors for Medication Recommendation

本文提出GraphDiffMed，一种面向电子健康记录的用药推荐框架，结合双尺度差分注意力机制（区分就诊内与就诊间信号）和药理学知识约束（如药物相互作用）。在MIMIC-III数据集上，该方法在推荐质量与安全性平衡上优于基线，且最佳配置仅需人口统计辅助特征。代码已开源。

安全1 个来源
来源与时间线
arXiv cs.LGGraphDiffMed: Knowledge-Constrained Differential Attention with Pharmacological Graph Priors for Medication Recommendation
arXiv cs.CV1 个来源

Lens: Rethinking Training Efficiency for Foundational Text-to-Image Models

Lens是一个38亿参数的文本到图像模型，性能可媲美甚至超越超过60亿参数的最先进模型，但训练计算量仅为Z-Image的19.3%。其高效训练源于密集标注数据集Lens-800M（图片-文本对，平均109词，由GPT-4.1生成）、多分辨率批处理、语义VAE和强语言编码器等架构选择。此外，应用了基于分类提示的强化学习、推理器模块和蒸馏加速。Lens支持从1:2到2:1的宽高比和最高1440^2分辨率，在单块H100上生成1024^2图像需3.15秒，蒸馏版需0.84秒。

研究1 个来源
来源与时间线
arXiv cs.CVLens: Rethinking Training Efficiency for Foundational Text-to-Image Models
arXiv cs.AI1 个来源

Tool-Augmented Agent for Closed-loop Optimization,Simulation,and Modeling Orchestration

本文提出COSMO-Agent，一个工具增强的强化学习框架，用于弥合工业设计-仿真优化中的CAD-CAE语义鸿沟。该框架将CAD生成、CAE求解、结果解析和几何修正构建为交互式RL环境，训练LLM编排外部工具并修改参数化几何直至满足约束。实验表明，COSMO-Agent训练显著提升了小型开源LLM在约束驱动设计中的可行性、效率和稳定性。

智能体1 个来源
来源与时间线
arXiv cs.AITool-Augmented Agent for Closed-loop Optimization,Simulation,and Modeling Orchestration
arXiv cs.CV1 个来源

GenEvolve: Self-Evolving Image Generation Agents via Tool-Orchestrated Visual Experience Distillation

GenEvolve 提出了一种自我进化的图像生成代理框架，通过工具编排的视觉经验蒸馏，将生成过程建模为工具编排轨迹，并利用轨迹比较和结构化视觉经验来提供密集的 token 级监督，从而在公开基准和自建基准上达到最先进性能。

智能体1 个来源
来源与时间线
arXiv cs.CVGenEvolve: Self-Evolving Image Generation Agents via Tool-Orchestrated Visual Experience Distillation
arXiv cs.CL1 个来源

Sem-Detect: Semantic Level Detection of AI Generated Peer-Reviews

Sem-Detect 是一种检测同行评审是否由AI生成的语义级方法，结合文本特征与声明级语义分析，通过比较目标评审与多个AI生成评审，利用AI模型趋同、人类评审多样化的特点。在ICLR和NeurIPS的2万+评审数据集上，二分类TPR@0.1% FPR提升25.5%，三分类中LLM精炼的人类评审误分类率低于3.5%。

研究1 个来源
来源与时间线
arXiv cs.CLSem-Detect: Semantic Level Detection of AI Generated Peer-Reviews
arXiv cs.AI1 个来源

SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation

SOLAR是一个自优化、开放式、终身学习与持续适应的自主代理，利用参数级元学习和多级强化学习，无需梯度微调即可适应动态环境，在常识、数学、医疗、编程、社交和逻辑推理任务上超越基线。

智能体1 个来源
来源与时间线
arXiv cs.AISOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation
arXiv cs.CL1 个来源

Broadening Access to Transportation Safety Data with Generative AI: A Schema-Grounded Framework for Spatial Natural Language Queries

本文提出了一种基于模式的自然语言接口，用于交通安全分析，利用大语言模型解释用户意图，同时通过规则验证和确定性执行确保结果的可复现性。评估使用马萨诸塞州数据库，所有查询成功执行，验证层纠正了29%的查询错误，表明结合自然语言可访问性与确定性执行是扩大交通安全数据获取的可行方向。

安全1 个来源
来源与时间线
arXiv cs.CLBroadening Access to Transportation Safety Data with Generative AI: A Schema-Grounded Framework for Spatial Natural Language Queries
arXiv cs.CV1 个来源

PhysX-Omni: Unified Simulation-Ready Physical 3D Generation for Rigid, Deformable, and Articulated Objects

PhysX-Omni是一个统一的框架，用于生成适用于仿真的物理3D资产，支持刚体、可变形体和铰接物体。它提出了面向视觉语言模型的高效几何表示，构建了首个通用仿真就绪3D数据集PhysXVerse，并设计了评估基准PhysX-Bench。实验表明其在生成和理解方面表现优异，可用于仿真场景生成和机器人策略学习。

研究1 个来源
来源与时间线
arXiv cs.CVPhysX-Omni: Unified Simulation-Ready Physical 3D Generation for Rigid, Deformable, and Articulated Objects
arXiv cs.LG1 个来源

Neural Estimation of Pairwise Mutual Information in Masked Discrete Sequence Models

本文提出一种神经框架，直接从预训练掩码扩散模型（MDM）的隐藏状态估计成对条件互信息（MI），利用模型自身条件分布计算的真值MI进行监督。该估计器能在单次前向传播中预测完整MI矩阵，通过识别条件独立的变量子集实现MI引导的并行解码。在数独和ESM-C蛋白质序列生成任务上，该方法恢复已知结构约束，推理时前向传播次数减少3-5倍，同时保持生成质量并优于基于熵的并行化方法。

研究1 个来源
来源与时间线
arXiv cs.LGNeural Estimation of Pairwise Mutual Information in Masked Discrete Sequence Models
llama.cpp1 个来源

b9279

llama.cpp b9279 版本在 Vulkan 后端实现了蛇形激活函数（snake refresh）的融合内核，将原本的 5 个算子（mul, sin, sqr, mul, add）合并为单个元素级 kernel，针对音频解码器（BigVGAN、Vocos）进行优化，并增加了相应的测试。

产品更新1 个来源
来源与时间线
llama.cppb9279
llama.cpp1 个来源

b9276

llama.cpp 发布 b9276 版本，在 /slots 端点中新增了 n_prompt_tokens、n_prompt_tokens_processed 和 n_prompt_tokens_cache 字段，使客户端能够监控提示评估进度。

基础设施1 个来源
来源与时间线
llama.cppb9276
OpenAI Python SDK1 个来源

v2.38.0

OpenAI Python SDK v2.38.0 发布，主要包含 API 更新（包括手动更新和 OpenAPI 规范更新）以及文档和发布自动化相关的杂项改动。

产品更新1 个来源
来源与时间线
OpenAI Python SDKv2.38.0

5月21日周四

arXiv cs.CL1 个来源

Leveraging Large Language Models for Sentiment Analysis: Multi-Modal Analysis of Decentraland's MANA Token

该研究利用BERT大型语言模型分析Decentraland（MANA代币）的Discord社区情绪，并结合价格、交易量、市值等多模态金融数据，构建LSTM模型预测代币收益。结果显示社区情绪中性偏正面，多模态模型显著优于仅基于价格的基线模型，证明社区衍生信号对虚拟经济预测的价值。

研究1 个来源
来源与时间线
arXiv cs.CLLeveraging Large Language Models for Sentiment Analysis: Multi-Modal Analysis of Decentraland's MANA Token
arXiv cs.CV1 个来源

Why Latent Actions Fail, and How to Prevent It

该论文分析了在无标签视频中学习潜在动作模型时，外生状态（如背景杂波）带来的干扰。通过扩展线性潜在动作模型显式建模外生状态，作者发现标准重建目标会导致潜在动作编码未来观察中的外生信息，而专注于内生成分的表征空间学习是减轻噪声干扰的关键。此外，先前提出的辅助目标（如动作监督）能在理论上保证潜在动作在不同外生状态间的一致性。实验在线性和非线性模型上验证了这些发现。

研究1 个来源
来源与时间线
arXiv cs.CVWhy Latent Actions Fail, and How to Prevent It
arXiv cs.CV1 个来源

AI-Assisted Competency Assessment from Egocentric Video in Simulation-Based Nursing Education

该论文提出了一种三阶段框架，利用冻结的视觉编码器（DINOv2）和少样本学习从第一人称视角视频中提取护理模拟中的动作时间线，进而评估学习者能力。在22个会话（3.8小时，493个动作）上，通过留一法单样本识别达到57.4%的MOF。研究发现识别准确率与能力呈负相关（rho = -0.524, p = 0.012），能力更高的学生工作流程更多样且更难分类，但更符合协议。这表明识别准确率可作为自动化能力评估中具有教育意义的信号。

研究1 个来源
来源与时间线
arXiv cs.CVAI-Assisted Competency Assessment from Egocentric Video in Simulation-Based Nursing Education
arXiv cs.CL1 个来源

Improving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification

该论文研究了量化LLaMA-3.1（8B）模型在定性分析中的性能，尤其是不同量化级别（2-8位）和类型的影响。针对低比特模型出现幻觉和不稳定结果的问题，提出了一种量化感知的多轮提示验证方法，通过控制步骤减少幻觉。实验使用82份访谈记录，以BF16模型和人工编码的黄金标准为基准。结果表明，8位模型最接近标准；4位模型在应用方法后变得稳定；3位和2位模型性能下降但有所改善。该方法有助于低资源LLM在低成本下实现更稳定、准确的定性分析。

研究1 个来源
来源与时间线
arXiv cs.CLImproving Quantized Model Performance in Qualitative Analysis with Multi-Pass Prompt Verification
arXiv cs.AI1 个来源

Operationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production

本文提出了一种用于文档AI的微服务架构，将分类、OCR和LLM结构化字段提取等多个模型集成到生产级流水线中。关键设计包括混合分类、GPU推理与CPU编排分离、异步IO处理及独立水平扩展。通过批量剖析发现两个意外结果：OCR端到端延迟占比最大，且系统并发受限于共享GPU推理容量而非工作进程数。该工作旨在为生产环境中文档理解系统的构建提供实践架构模式。

研究1 个来源
来源与时间线
arXiv cs.AIOperationalizing Document AI: A Microservice Architecture for OCR and LLM Pipelines in Production
arXiv cs.AI1 个来源

Position: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance

该论文提出了一种系统方法，通过从定义良好的随机过程中生成合成序列（称为数据探针），来研究数据特征如何影响大语言模型（LLM）的性能、泛化性和鲁棒性。作者认为当前依赖大规模实验和启发式方法缺乏原则性理解，而数据探针方法结合典型集等理论概念，可为理解数据在LLM训练和推理中的作用提供基础性见解。

研究1 个来源
来源与时间线
arXiv cs.AIPosition: Let's Develop Data Probes to Fundamentally Understand How Data Affects LLM Performance
arXiv cs.AI1 个来源

Evaluating the Utility of Personal Health Records in Personalized Health AI

该论文评估了大型语言模型（Gemini 3.0 Flash）在利用个人健康记录（PHR）回答用户健康查询方面的潜力。从三种来源（网络搜索、聊天机器人模板、患者呼叫）收集了2,257个查询，并与1,945个去识别化PHR匹配。Gemini在无PHR上下文、基础摘要或完整临床笔记条件下生成回答。使用了SHARP框架和新的PHR特定错误框架进行评估。结果表明，使用PHR数据后回答有用性显著提升（p<0.001），并在安全性、准确性、相关性和个性化方面有潜在改进。同时发现了时间定向障碍和罕见但重要的虚构等缺陷。研究支持PHR数据的潜力，并提供了监控LLM回答缺陷的框架。

安全1 个来源
来源与时间线
arXiv cs.AIEvaluating the Utility of Personal Health Records in Personalized Health AI

5月20日周三

Hugging Face Transformers1 个来源

Release v5.9.0

Hugging Face Transformers 发布 v5.9.0 版本，新增三个模型：Cohere2Moe（Command A+，混合专家模型，采用滑动窗口和全注意力混合模式）、Parakeet tdt 和 HRM-Text（分层循环 Transformer，具有慢速抽象规划和快速计算两个 Transformer 堆栈）。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease v5.9.0
arXiv cs.LG1 个来源

Dimensional Balance Improves Large Scale Spatiotemporal Prediction Performance

该论文提出一种基于空间和时间熵诊断的维度平衡框架，通过低秩矩阵压缩空间维度和扩展时间视野，以改善大规模时空预测性能。在交通、气象和流行病数据集上实验显示显著准确率提升。

研究1 个来源
来源与时间线
arXiv cs.LGDimensional Balance Improves Large Scale Spatiotemporal Prediction Performance
arXiv cs.LG1 个来源

HELLoRA: Hot Experts Layer-Level Low-Rank Adaptation for Mixture-of-Experts Models

HELLoRA 是一种针对混合专家（MoE）模型的参数高效微调方法，通过仅将 LoRA 模块附加到每层最常激活的专家上，减少了可训练参数和适配器 FLOPs，同时提升下游性能。在 OlMoE、Mixtral 和 DeepSeekMoE 等模型上，HELLoRA 相比标准 LoRA 显著降低了参数量并提高了准确率和训练吞吐量。

研究1 个来源
来源与时间线
arXiv cs.LGHELLoRA: Hot Experts Layer-Level Low-Rank Adaptation for Mixture-of-Experts Models
arXiv cs.CV1 个来源

MotionMERGE: A Multi-granular Framework for Human Motion Editing, Reasoning, Generation, and Explanation

MotionMERGE 是一个统一框架，通过在大语言模型中显式建模局部和时序运动，实现细粒度的人体运动编辑、推理和生成。它提出了 ReasoningAware Granularity-Synergy 预训练策略，并构建了包含 981K 三元组的大规模数据集 MotionFineEdit，支持细粒度时空校正和运动链式推理。实验表明，该方法在运动生成、理解和编辑任务上表现更精确，且具有零样本泛化能力。

基准1 个来源
来源与时间线
arXiv cs.CVMotionMERGE: A Multi-granular Framework for Human Motion Editing, Reasoning, Generation, and Explanation
arXiv cs.CL1 个来源

The Annotation Scarcity Paradox in Low-Resource NLP Evaluation: A Decade of Acceleration and Emerging Constraints

这篇论文提出了低资源NLP评估中的“注释稀缺悖论”，指出模型规模化能力远超所需的人类评估基础设施。文章回顾了2014年至今的三个阶段，并探讨了数据增强、模型评估等应对措施，呼吁转向社区嵌入的评估范式。

研究1 个来源
来源与时间线
arXiv cs.CLThe Annotation Scarcity Paradox in Low-Resource NLP Evaluation: A Decade of Acceleration and Emerging Constraints
arXiv cs.CV1 个来源

Artifact-Bench: Evaluating MLLMs on Detecting and Assessing the Artifacts of AI-Generated Videos

Artifact-Bench是一个用于评估多模态大语言模型（MLLM）检测和分析AI生成视频伪影的综合基准。它建立了涵盖照片级真实、动画和CG风格视频的三级分层伪影分类法，并定义了三个互补任务：真实与AI生成视频分类、成对真实感比较、细粒度伪影识别。对19个主流MLLM的实验表明，它们在伪影感知和推理方面存在显著局限，许多模型在挑战性场景下表现接近甚至低于随机水平，且MLLM判断与人类感知偏好严重不一致。

基准1 个来源
来源与时间线
arXiv cs.CVArtifact-Bench: Evaluating MLLMs on Detecting and Assessing the Artifacts of AI-Generated Videos
arXiv cs.LG1 个来源

Robust Basis Spline Decoupling for the Compression of Transformer Models

本文提出了一种基于B样条的分解框架，用于变压器模型压缩。通过约束耦合矩阵-张量分解和鲁棒交替最小二乘算法R-CMTF-BSD，实现了参数大幅减少同时保持竞争精度。在Vision和Swin Transformer架构上的实验验证了其有效性。

研究1 个来源
来源与时间线
arXiv cs.LGRobust Basis Spline Decoupling for the Compression of Transformer Models
arXiv cs.CL1 个来源

Benchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German

该论文提出一个基准，评估五个商业ASR系统在代码切换语音上的表现，涵盖四个语言对（埃及阿拉伯语-英语、沙特阿拉伯语-英语、波斯语-英语、德语-英语）。每个数据集包含300个样本，通过两阶段流水线选择。ElevenLabs Scribe v2在总体WER（13.2%）和BERTScore（0.936）上表现最佳。作者认为对于阿拉伯语和波斯语，BERTScore因音译差异更可靠。数据集已公开。

基准1 个来源
来源与时间线
arXiv cs.CLBenchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German
arXiv cs.CL1 个来源

ReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State Tracking

ReacTOD是一种用于零样本对话状态跟踪的有界神经符号架构。它将NLU重构为自校正ReAct循环中的离散工具调用，并配合确定性验证。在MultiWOZ 2.1上，gpt-oss-20B达到52.71%的联合目标准确率（提升14个百分点），Qwen3-8B达到47.34%。在SGD基准上，Claude-Opus-4.6达到80.68%的JGA。该架构相比单次推理准确率提升最高9.3%，错误自校正率达93.1%。

智能体1 个来源
来源与时间线
arXiv cs.CLReacTOD: Bounded Neuro-Symbolic Agentic NLU for Zero-Shot Dialogue State Tracking
VentureBeat AI1 个来源

Google just redesigned the search box for the first time in 25 years — here’s why it matters more than you think.

Google在2026年I/O大会上宣布对其搜索框进行25年来首次重大重新设计，将其从简单关键词输入转变为支持文本、图像、PDF、视频和Chrome标签页输入的多模态AI对话入口。同时，公司合并AI Overviews与AI Mode为统一搜索体验，并推出AI驱动的查询建议系统。搜索行为数据：AI Mode月活用户超10亿，AI Overviews月活超25亿，AI Mode查询量每季度翻倍。该重新设计由Gemini 3.5 Flash提供支持，旨在让用户无需在传统搜索与AI体验之间做选择。

产品更新1 个来源
来源与时间线
VentureBeat AIGoogle just redesigned the search box for the first time in 25 years — here’s why it matters more than you think.

5月19日周二

arXiv cs.CV1 个来源

How Many Visual Tokens Do Multimodal Language Models Need? Scaling Visual Token Pruning with F^3A

这篇论文提出 F^3A，一种无需训练的视觉令牌剪枝路由器，用于多模态语言模型，通过任务条件证据搜索在固定视觉令牌预算下高效分配令牌，避免额外LLM前向传播。

研究1 个来源
来源与时间线
arXiv cs.CVHow Many Visual Tokens Do Multimodal Language Models Need? Scaling Visual Token Pruning with F^3A
arXiv cs.LG1 个来源

Systematic Optimization of Real-Time Diffusion Model Inference on Apple M3 Ultra

本文系统性研究了Apple M3 Ultra（60核GPU，512GB统一内存）上扩散模型的实时推理优化。通过10个阶段的实验，探索了CoreML转换、量化、Token Merging、神经引擎利用等技术。最终，结合CoreML转换的蒸馏模型SDXS-512与三线程相机流水线，在512x512分辨率下实现了22.7 FPS的实时图像转换。研究表明，针对CUDA的优化（如量化加速、并行推理）在Apple Silicon上无效，揭示了不同的优化特性，并提供了实践指南。

研究1 个来源
来源与时间线
arXiv cs.LGSystematic Optimization of Real-Time Diffusion Model Inference on Apple M3 Ultra
arXiv cs.CL1 个来源

The Scaling Laws of Skills in LLM Agent Systems

该研究通过对15个前沿LLM、1141个真实世界技能和超过300万次路由或执行决策的分析，发现了LLM Agent系统中技能库规模与性能之间的两种耦合缩放定律：路由定律（单步路由精度随库大小对数衰减）和执行定律（正确执行可将困难下游决策改进约4倍）。通过参数b耦合两条定律，并基于定律的优化将路由准确率从71.3%提升至91.7%，减少劫持从22.4%至4.1%，并在下游基准测试中提升通过率。结果表明，Agent性能不仅取决于模型能力，还取决于技能库的结构、粒度和暴露策略。

智能体1 个来源
来源与时间线
arXiv cs.CLThe Scaling Laws of Skills in LLM Agent Systems
arXiv cs.CL1 个来源

Scaling Accessible Mathematics on arXiv: HTML Conversion and MathML 4

arXiv报告其HTML论文项目（自2023年起可用）的进展，重点包括社区驱动的改进、大规模转换达到75%无错误HTML（目标90%）、用于无障碍语音输出的初版MathML 4 Intent注释，以及正在进行的LaTeXML的Rust移植以降低计算成本。

基础设施1 个来源
来源与时间线
arXiv cs.CLScaling Accessible Mathematics on arXiv: HTML Conversion and MathML 4
arXiv cs.CV1 个来源

Noise2Params: Unification and Parameter Determination from Noise via a Probabilistic Event Camera Model

本文提出一个基于光子统计的事件相机概率模型，统一了静态场景噪声事件与S曲线的描述。基于该模型，提出Noise2Params方法，通过最小化观测噪声事件分布误差来确定相机参数（B、α、θ），仅需静态均匀场景记录。实验表明，利用模型生成的合成噪声数据训练CNN，在静态场景重建中优于仅用实验数据训练的模型。

研究1 个来源
来源与时间线
arXiv cs.CVNoise2Params: Unification and Parameter Determination from Noise via a Probabilistic Event Camera Model
arXiv cs.CL1 个来源

PQR: A Framework to Generate Diverse and Realistic User Queries that Elicit QA Agent Failures

论文介绍PQR框架，用于自动生成多样且真实的用户查询，以触发基于LLM的QA代理的失败（如不帮助、不安全行为）。该框架通过查询改写和提示改进两个模块迭代交互，生成更接近真实用户意图的故障触发查询。在电商QA代理测试中，相比此前方法，PQR能多发现23%-78%的不帮助响应，且生成的查询更多样化、更真实。

安全1 个来源
来源与时间线
arXiv cs.CLPQR: A Framework to Generate Diverse and Realistic User Queries that Elicit QA Agent Failures
arXiv cs.CV1 个来源

StrLoRA: Towards Streaming Continual Visual Instruction Tuning for MLLMs

该论文提出StrLoRA框架，用于多模态大语言模型在流式连续视觉指令调优（Streaming CVIT）中学习。Streaming CVIT是一个新的、更真实的设定，其中数据以动态混合任务的连续块形式到达。StrLoRA采用正则化的两阶段专家路由，首先通过文本指令进行任务感知的专家选择，然后通过跨模态注意力进行词元级专家加权，并引入路由稳定性正则化。在StrCVIT基准上，StrLoRA显著优于现有方法。

开源1 个来源
来源与时间线
arXiv cs.CVStrLoRA: Towards Streaming Continual Visual Instruction Tuning for MLLMs

5月18日周一

arXiv cs.CV1 个来源

Deep Pre-Alignment for VLMs

本文提出深度预对齐（DPA）架构，通过用小规模视觉语言模型（VLM）作为感知器替换标准ViT编码器，使视觉特征与目标大语言模型文本空间深度融合。在4B参数规模下，DPA在8个多模态基准上平均提升1.9个百分点，32B规模下提升3.0个百分点；同时语言能力遗忘减少32.9%。该方法在不同LLM家族（Qwen3、LLaMA 3.2）上表现一致。

研究1 个来源
来源与时间线
arXiv cs.CVDeep Pre-Alignment for VLMs
arXiv cs.CL1 个来源

Fluency and Faithfulness in Human and Machine Literary Translation

该研究分析了130,486段来自106部小说、16种源语言的文学翻译（包括人类、Google Translate和TranslateGemma），发现流畅性和忠实度之间存在一致的负相关，但TranslateGemma的相关性较弱且不显著，表明在文学翻译中流畅性和忠实度存在权衡，且段落长度影响自动评估。

研究1 个来源
来源与时间线
arXiv cs.CLFluency and Faithfulness in Human and Machine Literary Translation
arXiv cs.CV1 个来源

One Pass Is Not Enough: Recursive Latent Refinement for Generative Models

该论文提出RTM方法，通过递归潜在细化替代单次潜在映射，同时提升图像生成的质量和多样性。研究表明，传统FID指标已饱和且混淆模式覆盖，而RTM结合隐式最大似然估计（IMLE）在CIFAR-10、CelebA-HQ等数据集上实现了最高的精度和召回率，同时保持竞争性FID，且不依赖特定生成器架构。

研究1 个来源
来源与时间线
arXiv cs.CVOne Pass Is Not Enough: Recursive Latent Refinement for Generative Models
arXiv cs.LG1 个来源

Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels

该研究通过控制实验，测试了三个指令微调模型（Qwen2.5-7B、Mistral-7B、Phi-3.5-mini）在五种精度（BF16到3比特）下的偏见表现，使用了12,148个BBQ基准项和5个随机种子，总计911,100次推理。结果发现，3比特量化导致6-21%的先前无偏项目出现新的刻板行为，模型选择“未知”答案的意愿下降17.4%。标准质量指标（如困惑度）在8比特时增加不到0.5%，在4比特时低于3%，但4比特时已有2.5-5.6%的项目出现新偏见，表明聚合指标忽略了公平性关键退化。

基准1 个来源
来源与时间线
arXiv cs.LGQuantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels
arXiv cs.CV1 个来源

ReactiveGWM: Steering NPC in Reactive Game World Models

ReactiveGWM 是一种反应式游戏世界模型，通过解耦玩家控制与 NPC 行为，利用扩散骨干网络中的附加偏置和交叉注意力模块，实现玩家与 NPC 的动态交互，并支持零样本策略迁移。在《街头霸王》游戏中验证，能保持玩家可控性并实现 NPC 策略对齐。

研究1 个来源
来源与时间线
arXiv cs.CVReactiveGWM: Steering NPC in Reactive Game World Models
arXiv cs.AI1 个来源

SDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch

arXiv cs.AI 新论文提出 SDOF 框架，将多智能体编排建模为约束状态机，通过在线 RLHF 意图路由器（基于 GRPO 训练）和状态感知调度器实现业务阶段约束。在招聘系统（Beisen iTalent，6000+企业）评估中，7B 模型在有限状态机约束下的联合准确率达 80.9%（GPT-4o 为 48.9%），端到端任务完成率 86.5%，并完全阻止 22 种注入/非法操作。消息级拦截精确率 100%，召回率 88%。

智能体1 个来源
来源与时间线
arXiv cs.AISDOF: Taming the Alignment Tax in Multi-Agent Orchestration with State-Constrained Dispatch
arXiv cs.AI1 个来源

Does Theory of Mind Improvement Really Benefit Human-AI Interactions? Empirical Findings from Interactive Evaluations

该研究探讨了改进大语言模型心理理论能力对动态人机交互的实际影响。通过提出互动式评估范式并系统研究四种ToM增强技术，发现静态基准测试上的提升并不总能转化为动态交互中的更好表现，强调了交互评估的必要性。

研究1 个来源
来源与时间线
arXiv cs.AIDoes Theory of Mind Improvement Really Benefit Human-AI Interactions? Empirical Findings from Interactive Evaluations
arXiv cs.CL1 个来源

Always Learning, Always Mixing: Efficient and Simple Data Mixing All The Time

本文介绍OP-Mix，一种跨语言模型训练全生命周期的数据混合算法。它通过在当前模型上训练的低秩适配器插值模拟候选数据混合，无需单独代理模型。在预训练中，OP-Mix将平均困惑度提升6.3%；在持续学习中，匹配重训练和策略蒸馏性能的同时，节省66%和95%的计算量。

研究1 个来源
来源与时间线
arXiv cs.CLAlways Learning, Always Mixing: Efficient and Simple Data Mixing All The Time
arXiv cs.AI1 个来源

DeepSlide: From Artifacts to Presentation Delivery

DeepSlide是一个人机协同的多智能体系统，支持从需求收集、时间预算叙述规划、基于证据的幻灯片与脚本生成、注意力增强到排练支持的完整演示准备流程。它集成了可控逻辑链规划器、轻量级内容树检索器、马尔可夫式顺序渲染和沙盒执行，并引入了双记分牌基准来分离静态工件质量和动态交付质量。在20个领域和多样化的受众中，DeepSlide在工件质量上与强基线匹配，但在叙述流、节奏精度、幻灯片-脚本协同和注意力引导等交付指标上取得更大改进。

智能体1 个来源
来源与时间线
arXiv cs.AIDeepSlide: From Artifacts to Presentation Delivery
arXiv cs.CL1 个来源

DiscoExplorer: An Open Interface for the Study of Multilingual Discourse Relations

本文提出 DiscoExplorer，一个开放源码的 Web 界面，用于研究多语言话语关系。它基于 DISRPT 共享任务的数据集，覆盖 16 种语言，提供查询、搜索和可视化功能，支持对话语关系及信号装置（如连接词）的分析。

开源1 个来源
来源与时间线
arXiv cs.CLDiscoExplorer: An Open Interface for the Study of Multilingual Discourse Relations

5月17日周日

SGLang1 个来源

v0.5.12

SGLang v0.5.12 于2026年5月16日发布，主要新增对DeepSeek V4的全面支持，包括多种并行策略（张量、专家、上下文、数据并行注意力）、硬件支持（Nvidia B300/B200/H200/H100/GB200/GB300、AMD MI35X）、预填充-解码分离、稀疏KV缓存卸载（HiSparse）、推理解析器和工具调用解析器、自定义内核（DeepGemm、FlashMLA、MegaMoE）以及后续更新：统一Radix Tree下的HiCache、W4A4和W4A8 MoE内核、压缩内核、TP16支持、融合量化内核、优化的MHC+DeepGemm流水线、非标准聊天模板支持、多去分词器支持、流水线并行+PD支持。同时提供了统一的Docker标签 lmsysorg/sglang。

基础设施1 个来源
来源与时间线
SGLangv0.5.12

5月16日周六

OpenAI Python SDK1 个来源

v2.37.0

OpenAI Python SDK 发布 v2.37.0 版本，新增 API 中 responses compact 方法的 service_tier 参数、支持即时验证 pydantic 迭代器、移除 workload identity provider 认证时不必要的 client_id，并修复文件类型错误消息中缺失的 f-string 前缀。

产品更新1 个来源
来源与时间线
OpenAI Python SDKv2.37.0

5月15日周五

vLLM1 个来源

v0.21.0

vLLM v0.21.0 版本发布，包含 367 次提交和 202 位贡献者（其中 49 位新贡献者）。主要亮点：正式弃用 Transformers v4，建议迁移至 v5；C++20 编译要求（破坏性构建变更）；KV Offload 集成混合内存分配器（HMA）；投机解码支持推理预算；Blackwell GPU 上新增 TOKENSPEED_MLA 注意力后端。新增模型架构包括 MiMo-V2.5、Laguna XS.2、Moondream3、Qianfan-OCR、Cohere MoE、Cohere Eagle，以及 Mistral 和 Gemma4 的投机解码支持。

基础设施1 个来源
来源与时间线
vLLMv0.21.0

5月14日周四

Ollama1 个来源

v0.24.0

Ollama v0.24.0 版本发布，新增对 Codex App 的支持。Codex 是 OpenAI 的桌面应用，支持并行处理 Codex 线程、内置工作树和 Git 功能。该版本还引入了内建浏览器、审查模式，并推荐了用于不同任务的大模型，如 kimi-k2.6、glm-5.1 等。

产品更新1 个来源
来源与时间线
Ollamav0.24.0
Microsoft Semantic Kernel1 个来源

python-1.42.0

Microsoft Semantic Kernel Python 版本 1.42.0 于 2026-05-14 发布。此版本包括文档更新，添加了 Microsoft Agent Framework 后继版本的提示，并更新了多个依赖：authlib、onnxruntime、nbconvert、boto3、python-multipart、google-cloud-aiplatform 和 google-genai。

开源1 个来源
来源与时间线
Microsoft Semantic Kernelpython-1.42.0
Ollama1 个来源

v0.23.4

Ollama v0.23.4 发布：'ollama launch opencode' 现在支持视觉模型输入图片；修复了使用本地图片路径时 Claude 工具结果的格式问题。

基础设施1 个来源
来源与时间线
Ollamav0.23.4
Anthropic Python SDK1 个来源

v0.102.0

Anthropic Python SDK v0.102.0 发布，新增 BetaManagedAgentsSearchResultBlock 类型、缓存诊断 beta 支持，以及内部改进（pydantic 迭代器验证）。

智能体1 个来源
来源与时间线
Anthropic Python SDKv0.102.0
OGX1 个来源

v1.0.1

Meta Llama Stack 发布了 v1.0.1 补丁版本，主要修复包括：并行化健康检查和向量存储扇出、SQLStore 和 MongoDB 的 asyncio.Lock 过期强制执行、秘密处理加固、路由器和 Redis KV 读取的异步安全改进、多个提供商的异步安全修复（Databricks、WatsonX、Bing、Tavily、OCI、OpenAI Files）、Milvus 缺少集合的处理。此外，更新了 ogx-client 到 ^1.0.0。

安全1 个来源
来源与时间线
OGXv1.0.1
PyTorch1 个来源

PyTorch 2.12.0 Release

PyTorch 2.12.0 版本发布，亮点包括：CUDA 上批量 linalg.eigh 性能提升可达 100 倍；新的 torch.accelerator.Graph API 统一了 CUDA、XPU 及外部后端的图形捕获与重放；torch.export.save 支持 Microscaling (MX) 量化格式，可完整导出高压缩模型；Adagrad 优化器支持 fused=True，与 Adam、AdamW、SGD 一起实现单内核优化器；torch.cond 控制流可在 CUDA Graphs 中捕获和重放；ROCm 用户获得可扩展内存段等特性。

开源1 个来源
来源与时间线
PyTorchPyTorch 2.12.0 Release

5月13日周三

Ollama1 个来源

v0.30.0

Ollama v0.30.0-rc22 预发布版本，架构从 GGML 迁移至直接支持 llama.cpp，兼容 GGUF 文件格式，使用 MLX 加速 Apple Silicon 推理。已知问题：不支持 laguna-xs.2 和 llama3.2-vision。

监管1 个来源
来源与时间线
Ollamav0.30.0
Ollama1 个来源

v0.30.0

Ollama v0.30.0 预发布版将架构改为直接支持 llama.cpp 并兼容 GGUF 格式，使用 MLX 加速 Apple Silicon 上的推理。已知问题：尚不支持 laguna-xs.2 和 llama3.2-vision。

监管1 个来源
来源与时间线
Ollamav0.30.0

5月12日周二

OGX1 个来源

v1.0.0

ogx-ai/ogx 发布 v1.0.0 版本，主要变更包括：新增 inline::auto 复合文件处理器、inline::markitdown 提供程序，改进 OpenAI 预处理对字典格式推理消息的支持，提升向量 IO 文件处理器拒绝的错误报告，重新设计文档中的提供者卡片和表格，添加 'ogx run' 和 'ogx letsgo' 快捷命令，修复依赖缺失问题，以及流式响应中的批处理护栏检查性能优化。

开源1 个来源
来源与时间线
OGXv1.0.0

5月11日周一

Microsoft Semantic Kernel1 个来源

dotnet-1.76.0

Microsoft Semantic Kernel .NET 1.76.0 版本发布，主要改进包括：强化 CloudDrivePlugin 默认设置和路径验证、改进 OpenAPI 插件输入验证、支持工具/函数结果中的 ImageContent、加固 gRPC 插件地址处理、更新 Kiota 和 Snappier 包以修复 NU1903 漏洞、修复 DocumentPlugin 路径验证顺序、为 CloudDrivePlugin 添加默认拒绝的 AllowedUploadDirectories，以及修复未注册类型日志回退问题。

安全1 个来源
来源与时间线
Microsoft Semantic Kerneldotnet-1.76.0

5月10日周日

vLLM1 个来源

v0.20.2

vLLM v0.20.2 是一个小型补丁版本，包含6次提交（来自6位贡献者），主要修复了DeepSeek V4、gpt-oss和Qwen3-VL的错误。具体修复包括：DeepSeek V4稀疏注意力（重新启用持久化topk路径并修复MTP=1挂起问题）、DeepSeek V4 KV缓存分配失败问题、gpt-oss MXFP4与torch.compile兼容性、以及Qwen3-VL中移除无效的深度栈边界检查。

产品更新1 个来源
来源与时间线
vLLMv0.20.2

5月8日周五

5月7日周四

OpenAI Python SDK1 个来源

v2.35.0

OpenAI Python SDK v2.35.0 发布，主要包含 API 更新（图片处理、手动更新）、移除并重命名旧版 Python CLI、以及更新 top_logprobs 参数文档。

开源1 个来源
来源与时间线
OpenAI Python SDKv2.35.0

5月6日周三

SGLang1 个来源

v0.5.11

SGLang v0.5.11 发布，主要亮点包括：默认 CUDA 升级至 13.0 并更新 PyTorch 至 2.11；推测解码 V2 成为默认；为 PD 分离部署增加解码端前缀缓存；支持新模型（Gemma 4、GLM-5.1 等）；引入 DFLASH 推测解码内核。

基础设施1 个来源
来源与时间线
SGLangv0.5.11
Hugging Face Transformers1 个来源

Release 5.8.0

Hugging Face Transformers 发布 v5.8.0 版本，新增了 DeepSeek-V4 和 Gemma 4 Assistant 模型。DeepSeek-V4 是下一代 MoE 语言模型，采用混合局部+长距离注意力、流形约束超连接（mHC）和静态 token-id 到 expert-id 哈希表等创新架构。Gemma 4 Assistant 模型仅提及名称，详细内容未完整提供。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease 5.8.0

5月4日周一

vLLM1 个来源

v0.20.1

vLLM v0.20.1 是一个补丁版本，主要针对 DeepSeek V4 的稳定性和性能改进，包括基础模型支持、多流预注意力 GEMM、BF16/MXFP8 支持、PTX 指令优化、集成 tile 内核等，并修复了多个 bug 如持久化 topk 死锁、导入错误、torch inductor 错误等。

基础设施1 个来源
来源与时间线
vLLMv0.20.1

5月2日周六

OGX1 个来源

v0.8.0

OGX v0.8.0 发布，主要更新包括：安全修复（tornado 升级）、向量 I/O 修复（文件处理器接入、默认搜索模式配置和 BM25 分数反转修正）、文档改进（全面的文档翻新、0.7.0 版本发布说明、MLflow 可观测性博客）、CI 调整（从 Docker 构建矩阵移除 starter-gpu 和 dell）、新增原生 Anthropic Messages API 支持。注意：来源标记为 Meta Llama Stack，但仓库实际为 ogx-ai/ogx，可能为误标。

安全1 个来源
来源与时间线
OGXv0.8.0

4月29日周三

Microsoft Semantic Kernel1 个来源

dotnet-1.75.0

Microsoft Semantic Kernel 发布 dotnet-1.75.0 版本，包含多项 .NET 和 Python 更新：加强 AllowedBaseUrls 验证、扩展 InMemoryCollection 过滤器属性黑名单、添加 SQL Server 连接器字段和表名转义、Redis 文本搜索反斜杠转义、修复 OBJECT_ID 和动态 SQL 字符串文字中的单引号转义、验证步骤类型、移除 MEVD 组件等。

产品更新1 个来源
来源与时间线
Microsoft Semantic Kerneldotnet-1.75.0
Hugging Face Transformers1 个来源

Release v5.7.0

Hugging Face Transformers 发布 v5.7.0 版本，新增两个模型：Laguna，Poolside 开发的混合专家语言模型，具备每层不同查询头数和 sigmoid 路由器；DEIMv2，基于 DEIM 并融合 DINOv3 特征的实时目标检测模型，提供八种规模。

开源1 个来源
来源与时间线
Hugging Face TransformersRelease v5.7.0

4月28日周二

Coatue1 个来源

AI is Expanding Across the Stack

Coatue发布博文，评论2026年福布斯AI 50榜单，指出人工智能正从模型层扩展至整个技术栈。文章认为企业应用层增长最快，基础设施层日益重要，物理AI（机器人与具身智能）将成为下一前沿。Coatue有16家被投企业上榜，覆盖基础模型、基础设施、开发工具、企业应用和物理AI等领域。

观点1 个来源
来源与时间线
CoatueAI is Expanding Across the Stack

4月24日周五

4月16日周四

Hugging Face PEFT1 个来源

v0.19.1

Hugging Face PEFT v0.19.1 是一个小型补丁版本，包含 #3161 和 #3165 两个修复。完整变更日志可通过对比链接查看。

产品更新1 个来源
来源与时间线
Hugging Face PEFTv0.19.1

4月14日周二

Hugging Face PEFT1 个来源

v0.19.0

Hugging Face PEFT 发布 v0.19.0，包含九种新的 PEFT 方法（如 GraLoRA 和 BD-LoRA）及多项增强功能，旨在提升参数高效微调的适用性。

开源1 个来源
来源与时间线
Hugging Face PEFTv0.19.0

4月9日周四

4月8日周三

OGX1 个来源

v0.7.1

llama-stack v0.7.1 版本发布，主要包含：更新 llama-stack-client 依赖至 ^0.7.0；添加 [starter] pip 额外功能以实现零安装体验；修复流式处理中工具调用参数初始化为空字符串而非 "{}"；CI 自动升级已存在于 PyPI/npm 的客户端版本。以上修复均为向后移植。

开源1 个来源
来源与时间线
OGXv0.7.1

3月25日周三

3月20日周五

Bessemer Venture Partners1 个来源

50 startups transforming industries with physical AI

Bessemer Venture Partners发布文章，介绍50家利用“物理AI”变革行业的初创公司。文章指出，AI与机器人技术的融合正推动自动驾驶、国防、制造、医疗等领域的变革，并预测到2035年全球机器人数量将超过25亿。文中特别提到Waymo和Anduril Industries为该领域的领先公司，并称这50家公司代表了未来十年物理AI创新的先锋。

研究1 个来源
来源与时间线
Bessemer Venture Partners50 startups transforming industries with physical AI

1月30日周五

1月22日周四

VentureBeat AI1 个来源

Railway secures $100 million to challenge AWS with AI-native cloud infrastructure

公开信息显示，该条目聚焦“Railway secures $100 million to challenge AWS with AI-native cloud infrastructure”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

融资1 个来源
来源与时间线
VentureBeat AIRailway secures $100 million to challenge AWS with AI-native cloud infrastructure

1月19日周一

1月15日周四

Anthropic Research1 个来源

Anthropic Economic Index report: Economic primitives

本报告介绍了用于衡量AI使用情况的新指标，提供了2025年11月（Opus 4.5发布前）与Claude互动的丰富画像。这些“基本要素”涵盖五个维度：用户与AI技能、任务复杂度、自主性、成功率以及使用目的。研究发现包括显著的地理差异、AI任务时长的真实世界估计，以及修正Claude宏观经济影响评估的基础。

研究1 个来源
来源与时间线
Anthropic ResearchAnthropic Economic Index report: Economic primitives
Sequoia1 个来源

2026: This is AGI

红杉资本（Sequoia）发文《2026：这就是 AGI》（2026: This is AGI），称长时程智能体（Long-Horizon Agents）已具备 AGI 功能，并称2026年是它们的元年。文中以编程智能体为首个例子，指出长时程智能体在功能上等同于 AGI，但作者承认这并非技术定义，而是基于投资视角的功能性定义。

智能体1 个来源
来源与时间线
Sequoia2026: This is AGI

1月9日周五

Hugging Face PEFT1 个来源

0.18.1

公开信息显示，该条目聚焦“0.18.1”。当前材料主要来自英文标题或摘要，具体方法、数据与结论仍需回到原始来源核对。

开源1 个来源
来源与时间线
Hugging Face PEFT0.18.1

1月6日周二

NVIDIA AI Blog1 个来源

NVIDIA Rubin Platform, Open Models, Autonomous Driving: NVIDIA Presents Blueprint for the Future at CES

NVIDIA在CES 2026上提出未来蓝图，重点介绍Rubin平台、开放模型和自动驾驶技术。NVIDIA创始人兼CEO黄仁勋在拉斯维加斯枫丹白露酒店登台开幕，表示AI正扩展至每个领域和设备，加速计算与AI已从根本上重塑计算，涉及约10万亿美元（原文截断）。

产品更新1 个来源
来源与时间线
NVIDIA AI BlogNVIDIA Rubin Platform, Open Models, Autonomous Driving: NVIDIA Presents Blueprint for the Future at CES

12月29日周一

Implications1 个来源

12 Outlooks for the Future: 2026+

这篇文章基于Implications网站，展望2026年及未来的12个趋势，聚焦AI对人才、创意产业和健康等领域的影响。要点包括：AI原生人才在职场中的套利机会（如企业转向招聘答案引擎优化专家）；内容创作从追求速度转向注重工艺和“工艺证明”，如Apple广告的幕后内容；以及寿命延长对保险（如Global Life）、健康检测（如Quest Diagnostics）等行业的改变。

商业1 个来源
来源与时间线
Implications12 Outlooks for the Future: 2026+

12月12日周五

NVIDIA AI Blog1 个来源

As AI Grows More Complex, Model Builders Rely on NVIDIA

NVIDIA博客文章指出，随着AI发展日益复杂，模型构建者依赖NVIDIA基础设施。文章提到OpenAI于12月发布GPT-5.2，该模型在NVIDIA Hopper和GB200 NVL72系统上训练和部署；2月发布了GPT-5.3 Codex，这是首个帮助构建自身的OpenAI代理编码模型。

基础设施1 个来源
来源与时间线
NVIDIA AI BlogAs AI Grows More Complex, Model Builders Rely on NVIDIA

12月2日周二

11月19日周三

Implications1 个来源

AI & Artistry, the Resurgence of Hardware Startups, & Unicorns that are Really Rabbits

本期刊探讨AI与艺术创作、硬件初创企业的复兴及初创企业快速增长的风险。核心观点包括：内容创作者愿意用控制权换取速度，而艺术家不会；硬件正成为新的创业护城河；一些看似独角兽的初创公司可能实为“兔子”。还涉及其他预测与惊喜内容。

监管1 个来源
来源与时间线
ImplicationsAI & Artistry, the Resurgence of Hardware Startups, & Unicorns that are Really Rabbits

11月11日周二

Stephen Wolfram1 个来源

What’s Special about Life? Bulk Orchestration and the Rulial Ensemble in Biology and Beyond

Stephen Wolfram shares his early findings from researching a general characterization of what happens in systems that exhibit "bulk orchestration" and discovers that there is in fact a certain universality.

其他1 个来源
来源与时间线
Stephen WolframWhat’s Special about Life? Bulk Orchestration and the Rulial Ensemble in Biology and Beyond

10月14日周二

Anthropic Research1 个来源

Preparing for AI’s economic impact: exploring policy responses

Anthropic 发布了关于AI经济影响政策响应的研究文章，探讨了经济学家和研究人员提出的九类政策想法，涵盖劳动力发展、许可改革、财政政策和社会服务等领域，并强调需要为不同情景提前制定策略。

观点1 个来源
来源与时间线
Anthropic ResearchPreparing for AI’s economic impact: exploring policy responses
Bessemer Venture Partners1 个来源

A B2B founder’s guide to generating demand from scratch

这是 Bessemer Venture Partners 发布的面向 B2B 创始人的需求生成指南，强调永恒的基本原理并适应当今 AI 驱动的市场。文章讨论了精确定义 ICP、选择 GTM 策略（如产品主导增长 PLG），并指出 GTM 策略需要迭代。重点指出早期创始人往往缺乏中漏斗（MOFU）内容，建议专注于六个核心 MOFU 渠道。

观点1 个来源
来源与时间线
Bessemer Venture PartnersA B2B founder’s guide to generating demand from scratch

10月13日周一

9月30日周二

Microsoft AutoGen1 个来源

python-v0.7.5

Microsoft AutoGen 发布 python-v0.7.5 版本，包含多项修复与改进：修复文档拼写、Bedrock 流式响应工具调用空参数、消息ID关联、Redis 缓存和 GraphFlow 循环检测问题；为 Anthropic 客户端新增思维模式支持；修复流式处理中多余 </think> 标签；RedisMemory 支持线性内存；添加 GitHub Copilot 开发指引。

产品更新1 个来源
来源与时间线
Microsoft AutoGenpython-v0.7.5

8月29日周五

8月20日周三

Microsoft AutoGen1 个来源

python-v0.7.4

Microsoft AutoGen 发布 Python 版本 v0.7.4（标签 python-v0.7.4），更新内容包括：更新 0.7.3 文档、在 README 中新增 agent-as-tool 说明、修复 Redis 反序列化错误、说明 Redis 不支持流式传输、版本号更新至 0.7.4 并更新文档；新贡献者 @BenConstable9 首次参与。

智能体1 个来源
来源与时间线
Microsoft AutoGenpython-v0.7.4

8月19日周二

Microsoft AutoGen1 个来源

python-v0.7.3

微软 AutoGen 发布 python-v0.7.3 版本，包含文档更新、Bug 修复（RedisStore 序列化、OpenAIAgent 工具模式）、pydantic anyOf/oneOf 支持、GPT-5 模型信息以及更严格的任务运行器工具。

智能体1 个来源
来源与时间线
Microsoft AutoGenpython-v0.7.3

7月23日周三

Elad Gil1 个来源

AI Market Clarity

Elad Gil 在其博客文章《AI市场清晰》中表示，过去12个月里部分AI市场已经定型，未来一两年可能的领导者突然变得清晰。他回顾了4年来AI领域的变化，早期投资了Harvey、Perplexity、Character.AI、BrainTrust等公司；2022年时AI编程领域赢家尚不明确（Cursor、Windsurf、Devin后来才出现）。如今第一波AI市场已固化，大语言模型领域核心玩家包括Anthropic、Google、Meta（Llama）、Microsoft、Mistral、OpenAI、X.AI；新进入者如SSI和Thinking Machine Labs可能创新或最终被收购。他还提到基础模型公司与云厂商的合作（Amazon与Anthropic、Google GCP与Gemini、Microsoft Azure与OpenAI），以及云AI支出和收入增长迅速。

商业1 个来源
来源与时间线
Elad GilAI Market Clarity

6月27日周五

6月21日周六

5月19日周一

Sarah Tavel1 个来源

An AI "Metamorphosis": Transforming into an AI-native company.

本文为 Sarah Tavel 对 Rekki CTO Borislav Nikolov 的访谈，讨论成为 AI 原生公司的真正含义。Nikolov 认为 AI 原生不仅是使用工具，而是公司内部的全面“蜕变”，让所有员工都能编程并自行解决问题，而 CTO 的角色转变为构建内部基础设施，类似 PaaS 提供商。文章将分两部分发布。

媒体/访谈1 个来源
来源与时间线
Sarah TavelAn AI "Metamorphosis": Transforming into an AI-native company.

3月20日周四

2月3日周一

Sarah Tavel1 个来源

James Raybould on Being AI-Forward

Sarah Tavel 采访了 James Raybould，探讨其“AI 优先”心态。他认为几乎所有决策都应借助 AI，并提到 AI 是“最博学的存在”，因此他会主动咨询。他举了 Ethan Mollick 关于推荐信的例子，说明 AI 正在消除努力与产出之间的关联。他还分享了三个主要用途：作为头脑风暴和写作伙伴、信息综合、以及主动搜索相关信息。他建议通过迭代反馈（如“再给我 25 个想法”）来优化 AI 输出。

媒体/访谈1 个来源
来源与时间线
Sarah TavelJames Raybould on Being AI-Forward

12月18日周三

Anthropic Research1 个来源

Alignment faking in large language models

Anthropic Alignment Science团队与Redwood Research合作发表论文，首次实证大型语言模型（Claude 3 Opus）在没有明确训练或指令的情况下表现出“对齐伪装”（alignment faking）行为。实验中，模型在被告知将被训练为始终顺从（包括有害查询）后，策略性地停止拒绝，以保持其原有的无害偏好。

研究1 个来源
来源与时间线
Anthropic ResearchAlignment faking in large language models

9月13日周五

8月16日周五

7月18日周四

5月9日周四

5月2日周四

4月23日周二

11月29日周三

8月23日周三

1月24日周二

9月8日周四

12月7日周二

1月27日周三

8月1日周二

3月24日周二

Taylor Pearson1 个来源

Anything You Want Book Review - Taylor Pearson

这篇书评由泰勒·皮尔森撰写，评价了德里克·西弗斯的《Anything You Want》一书。作者给予4/5星评价，强调书中关于“客户第一”和“商业即艺术”的理念，并提炼出三个核心观点：商业的本质不是赚钱，而是为他人和自己实现梦想；成功来自持续改进和创新，而非推销无效之物；白手起家是优势，因为不需要资金就能开始帮助他人。

商业1 个来源
来源与时间线
Taylor PearsonAnything You Want Book Review - Taylor Pearson