AI 速递 2026-07-28
生成时间：2026/7/28 09:22:17（UTC: 2026-07-28T01:22:17.368Z）
AREX: Towards a Recursively Self-Improving Agent for Deep Research
Section titled “AREX: Towards a Recursively Self-Improving Agent for Deep Research”👍 142 · arXiv
Deep research requires agents to find answers that jointly satisfy multiple constraints. Discovering such answers is costly, whereas verifying a candidate can often be decomposed into tractable constr…
K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs
Section titled “K12-KGraph: A Curriculum-Aligned Knowledge Graph for Benchmarking and Training Educational LLMs”👍 57 · arXiv
Large language models are increasingly used in K-12 education, but existing benchmarks mainly test exam question answering rather than understanding how curriculum knowledge is structured and visually…
ReferTrack: Referring Then Tracking for Embodied Visual Tracking
Section titled “ReferTrack: Referring Then Tracking for Embodied Visual Tracking”👍 50 · arXiv
Embodied visual tracking (EVT) requires a mobile agent to continuously follow a specific target described in natural language using only onboard vision. While recent vision-language-action (VLA) polic…
Visual Contrastive Self-Distillation
Section titled “Visual Contrastive Self-Distillation”👍 44 · arXiv
On-policy self-distillation (OPSD) is promising as it removes the external teacher required by on-policy distillation (OPD), yet it still needs asymmetric information between teacher and student to en…
DataPrep-Bench: Benchmarking LLMs as Training Data Preparators
Section titled “DataPrep-Bench: Benchmarking LLMs as Training Data Preparators”👍 40 · arXiv
The quality of training data fundamentally determines the capabilities of large language models (LLMs), yet no unified benchmark exists to measure how well LLMs, agents, and data-centric workflows act…
LangChain langchain-fireworks==1.5.2
Section titled “LangChain langchain-fireworks==1.5.2”Changes since langchain-fireworks==1.5.1
release(fireworks): 1.5.2 (#39093) chore(model-profiles): refresh model profile data (#39092) chore(model-profiles): refresh model profile data (#39084) chore…
链接：https://github.com/langchain-ai/langchain/releases/tag/langchain-fireworks%3D%3D1.5.2
vLLM v0.26.0
Section titled “vLLM v0.26.0”vLLM v0.26.0 Release Notes
Section titled “vLLM v0.26.0 Release Notes”Highlights
Section titled “Highlights”This release features 411 commits from 212 contributors (61 new)!
- New Inkling model family with a full support stack: base modeling (#48799), …
链接：https://github.com/vllm-project/vllm/releases/tag/v0.26.0
Ollama v0.32.5
Section titled “Ollama v0.32.5”What’s Changed
Section titled “What’s Changed”- Fixed an MLX Metal bug that could reduce output quality for NVFP4 models, particularly Laguna.
Full Changelog: https://github.com/ollama/ollama/compare/v0.32.4…v0.32.5 …
链接：https://github.com/ollama/ollama/releases/tag/v0.32.5
CrewAI 1.15.7
Section titled “CrewAI 1.15.7”What’s Changed
Section titled “What’s Changed”Bug Fixes
Section titled “Bug Fixes”- Resolve registry skills through the runtime’s CrewAI+ client
- Recover from the GPT-5.6 tools + reasoning_effort 400
- Make tool calling work on the Responses API pa…
链接：https://github.com/crewAIInc/crewAI/releases/tag/1.15.7
OpenAI Codex CLI rust-v0.146.0-alpha.13
Section titled “OpenAI Codex CLI rust-v0.146.0-alpha.13”Release 0.146.0-alpha.13
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.146.0-alpha.13
Professor’s invisible prompt trap catches 32/35 students cheating with AI
Section titled “Professor’s invisible prompt trap catches 32/35 students cheating with AI”Article URL: https://www.techspot.com/news/113243-professor-invisible-prompt-trap-catches-32-students-cheating.html Comments URL: https://news.ycombinator.com/item?id=49074680 Points: 84
Comments: 77
Section titled “Comments: 77”Nvidia’s $750B in Deals Reignite Circular AI Fears
Section titled “Nvidia’s $750B in Deals Reignite Circular AI Fears”Article URL: https://www.bloomberg.com/news/articles/2026-07-27/nvidia-s-750-billion-deals-revive-fear-of-ai-circular-financing Comments URL: https://news.ycombinator.com/item?id=49071512 Points: 77
Comments: 73
Section titled “Comments: 73”Apple Will ‘Watch Everything Burn’ When the AI Bubble Bursts
Section titled “Apple Will ‘Watch Everything Burn’ When the AI Bubble Bursts”Article URL: https://www.macrumors.com/2026/07/27/ed-zitron-apple-watch-it-burn-ai-bubble-bursts/ Comments URL: https://news.ycombinator.com/item?id=49070427 Points: 233
Comments: 307
Section titled “Comments: 307”AI companies spend record sums on Washington lobbying
Section titled “AI companies spend record sums on Washington lobbying”Comments URL: https://news.ycombinator.com/item?id=49069939 Points: 253
Comments: 139
Section titled “Comments: 139”AI companies are shredding rare books
Section titled “AI companies are shredding rare books”https://xcancel.com/HedgieMarkets/status/2081534588485296565
Comments URL: https://news.ycombinator.com/item?id=49068738 Points: 737
Comments: 467
Section titled “Comments: 467”Elevated errors on Claude Opus 5
Section titled “Elevated errors on Claude Opus 5”Article URL: https://status.claude.com/incidents/mfdtrknpxghq Comments URL: https://news.ycombinator.com/item?id=49068029 Points: 97
Comments: 70
Section titled “Comments: 70”This July I Was Fired from Simple AI (A Deeply YC Company)
Section titled “This July I Was Fired from Simple AI (A Deeply YC Company)”Article URL: https://andys.blog/this-july-i-was-fired-from-simple-ai/ Comments URL: https://news.ycombinator.com/item?id=49059587 Points: 54
Comments: 171
Section titled “Comments: 171”The New AI Superpowers: Focus and Followthrough
Section titled “The New AI Superpowers: Focus and Followthrough”Article URL: https://www.rickmanelius.com/p/the-new-ai-superpowers-focus-and Comments URL: https://news.ycombinator.com/item?id=49057877 Points: 214