周报 2026-06-29 ~ 2026-07-05
生成时间：2026/7/5 12:58:57（UTC: 2026-07-05T04:58:57.585Z）
本周自动总结未启用或调用失败，以下为原始内容合并。
2026-06-29
Section titled “2026-06-29”生成时间：2026/6/29 10:11:45（UTC: 2026-06-29T02:11:45.664Z）
DanceOPD: On-Policy Generative Field Distillation
Section titled “DanceOPD: On-Policy Generative Field Distillation”👍 71 · arXiv
Modern image generation demands a single model that unifies diverse capabilities, including text-to-image (T2I), local editing, and global editing. However, these capabilities are rarely naturally ali…
In-Context World Modeling for Robotic Control
Section titled “In-Context World Modeling for Robotic Control”👍 51 · arXiv
Modern Vision-Language-Action (VLA) models often fail to generalize to novel setups, such as altered camera viewpoints or robot morphologies, because they are typically conditioned only on current obs…
OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning
Section titled “OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning”👍 46 · arXiv
Outcome-based reinforcement learning provides a stable optimization backbone for language agents, but its sparse trajectory-level rewards provide little guidance on which intermediate decisions should…
ShutterMuse: Capture-Time Photography Guidance with MLLMs
Section titled “ShutterMuse: Capture-Time Photography Guidance with MLLMs”👍 45 · arXiv
Real-world photography requires capture-time guidance for both camera framing and subject pose. Yet existing aesthetic cropping benchmarks mainly evaluate post-hoc crop prediction and overlook subject…
Qwen-Image-Agent: Bridging the Context Gap in Real-World Image Generation
Section titled “Qwen-Image-Agent: Bridging the Context Gap in Real-World Image Generation”👍 42 · arXiv
While text-to-image (T2I) models have achieved remarkable progress, they struggle with real-world requests that are often underspecified, implicit, or dependent on up-to-date knowledge. We identify th…
OpenClaw v2026.6.11-beta.2
Section titled “OpenClaw v2026.6.11-beta.2”Highlights
Section titled “Highlights”- More capable channel control: Slack relay mode, native Mattermost
/oc_queue, and per-DM model overrides make channel operations easier to automate and tune. (#94707, #95546, #…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.6.11-beta.2
CrewAI 1.15.1
Section titled “CrewAI 1.15.1”What’s Changed
Section titled “What’s Changed”Features
Section titled “Features”- Initialize Git repositories for generated projects (#6364)
- Require explicit CrewAI project definitions (#6358)
- Open deployment page after CLI deploy (#6343)
…
链接：https://github.com/crewAIInc/crewAI/releases/tag/1.15.1
OpenAI Codex CLI rust-v0.143.0-alpha.29
Section titled “OpenAI Codex CLI rust-v0.143.0-alpha.29”Release 0.143.0-alpha.29
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.143.0-alpha.29
Ford rehires ‘gray beard’ engineers after AI falls short
Section titled “Ford rehires ‘gray beard’ engineers after AI falls short”“Mistakenly we thought that by just introducing artificial intelligence … that would produce a high-quality product.”
Why Wall Street thinks US memory maker Micron is the next Nvidia
Section titled “Why Wall Street thinks US memory maker Micron is the next Nvidia”Eager to find more public AI-related companies that may do as well as Nvidia, Wall Street investors think they’ve found a winner with Micron.
SoftBank’s CEO isn’t the only one with questions about Elon Musk’s orbital data center hype
Section titled “SoftBank’s CEO isn’t the only one with questions about Elon Musk’s orbital data center hype”Not everyone is buying Elon Musk’s vision for orbital data centers.
Apple Vision Pro exec is reportedly leaving for OpenAI
Section titled “Apple Vision Pro exec is reportedly leaving for OpenAI”Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving the company to join OpenAI’s hardware team.
The fittest founder in the room got cancer. Here’s how he used AI to fight back.
Section titled “The fittest founder in the room got cancer. Here’s how he used AI to fight back.”When confronted with cancer, Connor Christou fed everything tied tied to his regime — blood results, scan data, wearable output, journal entries — into Claude.
Asian AI startups launch Mythos-like models as Anthropic’s export ban drags on
Section titled “Asian AI startups launch Mythos-like models as Anthropic’s export ban drags on”New models are launching in Asia that promise Mythos-like capabilities without fear of an export ban. U.S. AI labs may never recover this enormous market.
Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies
Section titled “Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies”Over 100 companies and government agencies are reportedly authorized to use Mythos 5, including their non-American employees.
OpenAI limits GPT-5.6 rollout after government request, says restrictions shouldn’t be the norm
Section titled “OpenAI limits GPT-5.6 rollout after government request, says restrictions shouldn’t be the norm”“We don’t believe this kind of government access process should become the long-term default,” says OpenAI. “It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.”
2026-06-30
Section titled “2026-06-30”生成时间：2026/6/30 10:06:48（UTC: 2026-06-30T02:06:48.813Z）
In-Context World Modeling for Robotic Control
Section titled “In-Context World Modeling for Robotic Control”👍 57 · arXiv
Modern Vision-Language-Action (VLA) models often fail to generalize to novel setups, such as altered camera viewpoints or robot morphologies, because they are typically conditioned only on current obs…
OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning
Section titled “OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning”👍 49 · arXiv
Outcome-based reinforcement learning provides a stable optimization backbone for language agents, but its sparse trajectory-level rewards provide little guidance on which intermediate decisions should…
Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs
Section titled “Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs”👍 43 · arXiv
We introduce an axiomatic evaluation framework for latent thought representations in LLMs, comprising metrics that are independent of downstream benchmark scores and reveal representational failures t…
The Verification Horizon: No Silver Bullet for Coding Agent Rewards
Section titled “The Verification Horizon: No Silver Bullet for Coding Agent Rewards”👍 43 · arXiv
A classical intuition holds that verifying a solution is easier than producing one. For today’s coding agents, this intuition is being inverted: as foundation models develop stronger reasoning capabil…
PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation
Section titled “PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation”👍 41 · arXiv
Video generation models have emerged as a promising paradigm for embodied world simulation. However, both general-domain video generators and robot-specific data fine-tuned models can still produce ph…
OpenClaw v2026.6.11-beta.2
Section titled “OpenClaw v2026.6.11-beta.2”2026.6.11
Section titled “2026.6.11”Highlights
Section titled “Highlights”- More capable channel control: Slack relay mode, native Mattermost
/oc_queue, and per-DM model overrides make channel operations easier to automate and tune. (#947…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.6.11-beta.2
LangChain langchain-openrouter==0.2.5
Section titled “LangChain langchain-openrouter==0.2.5”Changes since langchain-openrouter==0.2.4
release(openrouter): 0.2.5 (#38553) fix(openrouter): deduplicate repeated finish metadata (#38552) fix(openrouter): strip Responses reasoning IDs (#38383)…
链接：https://github.com/langchain-ai/langchain/releases/tag/langchain-openrouter%3D%3D0.2.5
vLLM v0.24.0
Section titled “vLLM v0.24.0”vLLM v0.24.0 Release Notes
Section titled “vLLM v0.24.0 Release Notes”Highlights
Section titled “Highlights”This release features 571 commits from 256 contributors (77 new)!
- MiniMax-M3: Added support for the new MiniMax-M3 model (#45381), with a …
链接：https://github.com/vllm-project/vllm/releases/tag/v0.24.0
Ollama v0.30.12-rc0
Section titled “Ollama v0.30.12-rc0”What’s Changed
Section titled “What’s Changed”- tools: ignore braces inside JSON strings when detecting tool call end by @aditya-786 in https://github.com/ollama/ollama/pull/16937
- mlx: bump dependency by @dhiltgen in https:/…
链接：https://github.com/ollama/ollama/releases/tag/v0.30.12-rc0
OpenAI Codex CLI rust-v0.142.4
Section titled “OpenAI Codex CLI rust-v0.142.4”Chores
Section titled “Chores”- No user-facing changes were identified for this release.
Changelog
Section titled “Changelog”Full Changelog: https://github.com/openai/codex/compare/rust-v0.142.3…rust-v0.142.4
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.142.4
Micro-Agent: Beat Frontier Models with Collaboration Inside Model API
Section titled “Micro-Agent: Beat Frontier Models with Collaboration Inside Model API”Article URL: https://vllm.ai/blog/2026-06-29-micro-agent-frontier-models Comments URL: https://news.ycombinator.com/item?id=48722802 Points: 53
Comments: 18
Section titled “Comments: 18”Working With AI: A concrete example
Section titled “Working With AI: A concrete example”Article URL: https://htmx.org/essays/working-with-ai/ Comments URL: https://news.ycombinator.com/item?id=48720064 Points: 90
Comments: 31
Section titled “Comments: 31”Tidal AI Policy
Section titled “Tidal AI Policy”Article URL: https://tidal.com/ai-policy Comments URL: https://news.ycombinator.com/item?id=48718840 Points: 292
Comments: 321
Section titled “Comments: 321”Anthropic CEO: Open-Source AI is getting dangerous (2023)
Section titled “Anthropic CEO: Open-Source AI is getting dangerous (2023)”Article URL: https://xcancel.com/coinbureau/status/2071330294452666695 Comments URL: https://news.ycombinator.com/item?id=48716750 Points: 51
Comments: 24
Section titled “Comments: 24”Herdr: Agent multiplexer that lives in your terminal
Section titled “Herdr: Agent multiplexer that lives in your terminal”Article URL: https://github.com/ogulcancelik/herdr Comments URL: https://news.ycombinator.com/item?id=48714802 Points: 145
Comments: 90
Section titled “Comments: 90”AI boom risks global financial crash, warn central bankers
Section titled “AI boom risks global financial crash, warn central bankers”Article URL: https://www.telegraph.co.uk/business/2026/06/28/ai-boom-risks-global-financial-crash-central-bankers-warn/ Comments URL: https://news.ycombinator.com/item?id=48713697 Points: 155
Comments: 203
Section titled “Comments: 203”Better Images of AI
Section titled “Better Images of AI”Article URL: https://betterimagesofai.org/ Comments URL: https://news.ycombinator.com/item?id=48713051 Points: 55
Comments: 30
Section titled “Comments: 30”We need tech news sources which exclude AI
Section titled “We need tech news sources which exclude AI”Its now clear that we need to preserve tech press for non AI related things. Techmeme for example is now completely overrun with AI stories.HN is getting closer to that every day.If AI kickback deals, phony new model ratings, high RAM prices and your surprise at how you think you coded something wit
2026-07-01
Section titled “2026-07-01”生成时间：2026/7/1 10:11:31（UTC: 2026-07-01T02:11:31.610Z）
Agentic Abstention: Do Agents Know When to Stop Instead of Act?
Section titled “Agentic Abstention: Do Agents Know When to Stop Instead of Act?”👍 121 · arXiv
LLM agents are expected to act over multiple turns, using search, browsing interfaces, and terminal tools to complete user goals. Yet not every goal is well specified or achievable in the available en…
LiveEdit: Towards Real-Time Diffusion-Based Streaming Video Editing
Section titled “LiveEdit: Towards Real-Time Diffusion-Based Streaming Video Editing”👍 72 · arXiv
Streaming video editing has made rapid progress, yet practical deployment is still limited by two core issues: maintaining stable backgrounds and non-edited regions over time, and achieving the low la…
Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent
Section titled “Scaling the Horizon, Not the Parameters: Reaching Trillion-Parameter Performance with a 35B Agent”👍 67 · arXiv
We introduce Agents-A1, a 35B Mixture-of-Experts Agentic Model that reaches trillion-parameter-level performance by scaling the agent horizon. We investigate agent-horizon scaling from two perspective…
TUA-Bench: A Benchmark for General-Purpose Terminal-Use Agents
Section titled “TUA-Bench: A Benchmark for General-Purpose Terminal-Use Agents”👍 44 · arXiv
As large language models and harness frameworks continue to advance, agents operating in terminals are increasingly capable of performing a broader range of general computer-use tasks beyond coding. H…
ReFreeKV: Towards Threshold-Free KV Cache Compression
Section titled “ReFreeKV: Towards Threshold-Free KV Cache Compression”👍 43 · arXiv
To reduce memory consumption during LLM inference, a handful of methods have been proposed for KV cache pruning. While these techniques can accomplish lossless memory reduction on many datasets, they …
OpenClaw v2026.6.11
Section titled “OpenClaw v2026.6.11”We heard the feedback. v2026.6.11 focuses on the rough edges that make OpenClaw feel less dependable, with fixes for misplaced replies, stuck sends, reconnects, model setup failures, and safer admin d…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.6.11
LangChain langchain-openrouter==0.2.5
Section titled “LangChain langchain-openrouter==0.2.5”Changes since langchain-openrouter==0.2.4
release(openrouter): 0.2.5 (#38553) fix(openrouter): deduplicate repeated finish metadata (#38552) fix(openrouter): strip Responses reasoning IDs (#38383)…
链接：https://github.com/langchain-ai/langchain/releases/tag/langchain-openrouter%3D%3D0.2.5
vLLM v0.24.0
Section titled “vLLM v0.24.0”vLLM v0.24.0 Release Notes
Section titled “vLLM v0.24.0 Release Notes”Highlights
Section titled “Highlights”This release features 571 commits from 256 contributors (77 new)!
- MiniMax-M3: Added support for the new MiniMax-M3 model (#45381), with a …
链接：https://github.com/vllm-project/vllm/releases/tag/v0.24.0
Ollama v0.31.1
Section titled “Ollama v0.31.1”Faster Gemma 4 on Apple Silicon
Section titled “Faster Gemma 4 on Apple Silicon”链接：https://github.com/ollama/ollama/releases/tag/v0.31.1
CrewAI 1.15.2a1
Section titled “CrewAI 1.15.2a1”What’s Changed
Section titled “What’s Changed”Features
Section titled “Features”- Repoint template commands to crewAIInc-fde org
- Support inline skill definitions
- Define stream frame protocol for flows
- Add type tool and app in CrewDefinition -…
链接：https://github.com/crewAIInc/crewAI/releases/tag/1.15.2a1
OpenAI Codex CLI rust-v0.142.5
Section titled “OpenAI Codex CLI rust-v0.142.5”Bug Fixes
Section titled “Bug Fixes”- Prevented full Responses WebSocket request payloads from being written to trace logs. (#30771)
Changelog
Section titled “Changelog”Full Changelog: https://github.com/openai/codex/compare/rust-v0.142.4…ru…
链接：https://github.com/openai/codex/releases/tag/rust-v0.142.5
Wayve launches $85M employee tender offer at $8.5B valuation
Section titled “Wayve launches $85M employee tender offer at $8.5B valuation”Wayve’s offering is part of a growing trend of AI startups using employee tenders as a strategic tool to attract and retain talent.
OpenClaw is finally available on Android and iOS
Section titled “OpenClaw is finally available on Android and iOS”The free open source agentic program is finally invading your phone.
The DeepMind trio who built a poker AI are now making money for quant hedge funds
Section titled “The DeepMind trio who built a poker AI are now making money for quant hedge funds”EquiLibre Technologies, a Prague-based AI lab founded by three ex-DeepMind researchers, is now valued at more than $500 million.
Google introduces a faster, cheaper image generator with Nano Banana 2 Lite
Section titled “Google introduces a faster, cheaper image generator with Nano Banana 2 Lite”Google is updating its image generator to make it faster and cheaper, making it a more useful tool for creators looking to make AI content.
Nvidia competitor Etched hits $5B valuation, $1B in sales for AI chip
Section titled “Nvidia competitor Etched hits $5B valuation, $1B in sales for AI chip”Nvidia AI chip competitor Etched says it has already booked $1 billion under contract for the inference systems powered by its chip.
Anthropic launches Claude Sonnet 5 as a cheaper way to run agents
Section titled “Anthropic launches Claude Sonnet 5 as a cheaper way to run agents”Anthropic’s Claude Sonnet 5 brings stronger agentic capabilities, lower pricing, and improved safety, positioning the model as a cheaper alternative to Opus, GPT-5.5, and Gemini Pro.
Acti puts AI agents directly into your smartphone keyboard
Section titled “Acti puts AI agents directly into your smartphone keyboard”Acti is betting the smartphone keyboard is the next home for AI assistants. The startup’s new keyboard for iOS and Android works across apps and lets users create custom AI-powered shortcuts using natural language.
Anthropic’s Claude Science bets on workflow, not a new model, to win over scientists
Section titled “Anthropic’s Claude Science bets on workflow, not a new model, to win over scientists”Anthropic’s Claude Science is a workbench that gives scientists one environment to do computational research, saving them from the need to bounce between databases, pipelines, and tools.
2026-07-02
Section titled “2026-07-02”生成时间：2026/7/2 10:03:51（UTC: 2026-07-02T02:03:51.040Z）
Orca: The World is in Your Mind
Section titled “Orca: The World is in Your Mind”👍 177 · arXiv
We introduce Orca, an initial instantiation of a general world foundation model. Orca learns a unified world latent space from multimodal world signals and exposes it through multimodal readout interf…
Dockerless: Environment-Free Program Verifier for Coding Agents
Section titled “Dockerless: Environment-Free Program Verifier for Coding Agents”👍 86 · arXiv
Program verifiers play a central role in training coding agents, including selecting trajectories for supervised fine-tuning (SFT) and providing rewards for reinforcement learning (RL). Standard execu…
DOPD: Dual On-policy Distillation
Section titled “DOPD: Dual On-policy Distillation”👍 76 · arXiv
On-policy distillation (OPD) offers superior capacity transfer by supervising student-sampled trajectories with dense token-level signals. To furnish high-quality supervision sources and thereby eleva…
BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decoding
Section titled “BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decoding”👍 67 · arXiv
Speculative decoding accelerates inference by using a lightweight draft model to generate candidate tokens in parallel, and are then verified by the target model, enabling lossless acceleration. Recen…
Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models
Section titled “Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models”👍 41 · arXiv
Embodied Vision-Language-Action (VLA) models are typically obtained by fine-tuning powerful pretrained VLMs on robotics data, yet it is unclear how much commonsense and factual knowledge they retain a…
OpenClaw v2026.6.11
Section titled “OpenClaw v2026.6.11”We heard the feedback. v2026.6.11 focuses on the rough edges that make OpenClaw feel less dependable, with fixes for misplaced replies, stuck sends, reconnects, model setup failures, and safer admin d…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.6.11
Ollama v0.31.1
Section titled “Ollama v0.31.1”Faster Gemma 4 on Apple Silicon
Section titled “Faster Gemma 4 on Apple Silicon”链接：https://github.com/ollama/ollama/releases/tag/v0.31.1
CrewAI 1.15.2a2
Section titled “CrewAI 1.15.2a2”What’s Changed
Section titled “What’s Changed”Features
Section titled “Features”- Add aiobotocore to the bedrock extra
- Document flow agent options
- Add text helper to flow skill example
- Add text helper for flow CEL prompts
- Add streaming docs…
链接：https://github.com/crewAIInc/crewAI/releases/tag/1.15.2a2
OpenAI Codex CLI rust-v0.143.0-alpha.33
Section titled “OpenAI Codex CLI rust-v0.143.0-alpha.33”Release 0.143.0-alpha.33
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.143.0-alpha.33
SpaceX has an AI device prototype, and it sure sounds phone-ish
Section titled “SpaceX has an AI device prototype, and it sure sounds phone-ish”SpaceX reportedly showed investors a “handset-like” AI device before going public. It could be another signal SpaceX wants to expand into wireless.
Ashton Kutcher leaving Sound Ventures to launch new VC firm with Morgan Beller
Section titled “Ashton Kutcher leaving Sound Ventures to launch new VC firm with Morgan Beller”Sound built its reputation on concentrated, high-conviction bets in category-leading AI labs, while Kutcher’s new fund appears to be chasing the layer underneath those companies — the infrastructure and energy that power them.
Cloudflare’s new policy pushes AI companies to pay for publishers’ content
Section titled “Cloudflare’s new policy pushes AI companies to pay for publishers’ content”Cloudflare is giving AI companies until September 15 to separate web crawlers used for search from those used for AI training and agents, or risk being blocked by default on many publisher sites.
Venice AI becomes a unicorn with $65M Series A as its privacy-first AI platform takes off
Section titled “Venice AI becomes a unicorn with $65M Series A as its privacy-first AI platform takes off”Venice AI is already profitable, with annualized run-rate revenues of over $70 million, CEO Erik Voorhees said.
Gemini Spark, Google’s agentic assistant, is now available on Mac
Section titled “Gemini Spark, Google’s agentic assistant, is now available on Mac”Google’s 24/7 agentic assistant, Gemini Spark, comes to Mac alongside other improvements, like real-time tracking and support for more apps.
Builders Stage agenda revealed: Practical strategies for scaling startups at TechCrunch Disrupt 2026
Section titled “Builders Stage agenda revealed: Practical strategies for scaling startups at TechCrunch Disrupt 2026”The Builders Stage is returning to TechCrunch Disrupt 2026, bringing together 10,000+ founders, startup operators, and investors for practical conversations. and Q&A on what it takes to build and scale successful companies. Register now to save up to $330.
Meta, like SpaceX, looks to turn excess AI compute into cash
Section titled “Meta, like SpaceX, looks to turn excess AI compute into cash”Meta is developing plans for a cloud infrastructure business, selling access to AI compute power and models. The move would pit it against the big cloud providers like Amazon Web Services, Google Cloud, and Microsoft Azure.
The ‘Father of the Internet’ is finally retiring
Section titled “The ‘Father of the Internet’ is finally retiring”Vinton Cerf, one of the creators of the protocols underlying the internet, will step down as Google’s chief internet evangelist next week.
2026-07-03
Section titled “2026-07-03”生成时间：2026/7/3 09:47:58（UTC: 2026-07-03T01:47:58.485Z）
Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models
Section titled “Does VLA Even Know the Basics? Measuring Commonsense and World Knowledge Retention in Vision-Language-Action Models”👍 54 · arXiv
Embodied Vision-Language-Action (VLA) models are typically obtained by fine-tuning powerful pretrained VLMs on robotics data, yet it is unclear how much commonsense and factual knowledge they retain a…
PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception
Section titled “PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception”👍 34 · arXiv
We introduce PerceptionRubrics, a rubric-based evaluation framework that addresses the gap between saturated benchmark scores and real-world brittleness. Shifting evaluation from holistic semantic mat…
SkillHone: A Harness for Continual Agent Skill Evolution Through Persistent Decision History
Section titled “SkillHone: A Harness for Continual Agent Skill Evolution Through Persistent Decision History”👍 24 · arXiv
Agent skills extend language-model agents with task-specific procedures, scripts, and references, but the tasks and environments they target continually change. Existing methods improve skills in boun…
ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving
Section titled “ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving”👍 19 · arXiv
In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models this is inco…
TurboServe: Serving Streaming Video Generation Efficiently and Economically
Section titled “TurboServe: Serving Streaming Video Generation Efficiently and Economically”👍 19 · arXiv
Streaming video generation is emerging as a new serving workload in which users interact with long-lived sessions that generate video progressively, chunk by chunk. Unlike offline video generation or …
OpenClaw v2026.7.1-beta.1
Section titled “OpenClaw v2026.7.1-beta.1”Highlights
Section titled “Highlights”- OpenAI GPT-5.6 support: OpenClaw now recognizes the GPT-5.6 model family across catalog, capability, and runtime selection paths. (#98333) Thanks @steipete-oai.
- **External har…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.7.1-beta.1
CrewAI 1.15.2a2
Section titled “CrewAI 1.15.2a2”What’s Changed
Section titled “What’s Changed”Features
Section titled “Features”- Add aiobotocore to the bedrock extra
- Document flow agent options
- Add text helper to flow skill example
- Add text helper for flow CEL prompts
- Add streaming docs…
链接：https://github.com/crewAIInc/crewAI/releases/tag/1.15.2a2
Goose v1.41.0
Section titled “Goose v1.41.0”✨ Features
Section titled “✨ Features”- iFlytek Spark and Astron MaaS providers (#9837)
--editsession flag to edit conversation before forking (#9799)- Fireworks AI declarative provider (#9990)
- OpenRouter request pa…
链接：https://github.com/aaif-goose/goose/releases/tag/v1.41.0
OpenAI Codex CLI rust-v0.143.0-alpha.34
Section titled “OpenAI Codex CLI rust-v0.143.0-alpha.34”Release 0.143.0-alpha.34
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.143.0-alpha.34
The short leash AI coding method for beating Fable
Section titled “The short leash AI coding method for beating Fable”Article URL: https://blog.okturtles.org/2026/07/short-leash-ai-method/ Comments URL: https://news.ycombinator.com/item?id=48766026 Points: 63
Comments: 63
Section titled “Comments: 63”Claude-real-video － any LLM can watch a video
Section titled “Claude-real-video － any LLM can watch a video”Article URL: https://github.com/HUANGCHIHHUNGLeo/claude-real-video Comments URL: https://news.ycombinator.com/item?id=48766005 Points: 81
Comments: 28
Section titled “Comments: 28”Claude’s AskUserQuestion: “No response after 60s – continued without an answer”
Section titled “Claude’s AskUserQuestion: “No response after 60s – continued without an answer””Article URL: https://github.com/anthropics/claude-code/issues/73125 Comments URL: https://news.ycombinator.com/item?id=48765630 Points: 54
Comments: 59
Section titled “Comments: 59”Weird Al Yankovic Pulled Out of AI Ad Deal: ‘I Can’t Be the Poster Boy for AI’
Section titled “Weird Al Yankovic Pulled Out of AI Ad Deal: ‘I Can’t Be the Poster Boy for AI’”Article URL: https://variety.com/2026/biz/news/weird-al-yankovic-rejected-ai-commercial-money-offer-1236800794/ Comments URL: https://news.ycombinator.com/item?id=48764326 Points: 59
Comments: 33
Section titled “Comments: 33”No LLM Code in Dependencies
Section titled “No LLM Code in Dependencies”Article URL: https://joeyh.name/blog/entry/no_LLM_code_in_dependencies/ Comments URL: https://news.ycombinator.com/item?id=48762008 Points: 114
Comments: 97
Section titled “Comments: 97”AI can’t be listed as inventor on patent applications, Japan’s top court rules
Section titled “AI can’t be listed as inventor on patent applications, Japan’s top court rules”Article URL: https://japannews.yomiuri.co.jp/science-nature/technology/20260306-314930/ Comments URL: https://news.ycombinator.com/item?id=48761536 Points: 360
Comments: 190
Section titled “Comments: 190”AI fake news complaining about how AI fake news is the death of real news
Section titled “AI fake news complaining about how AI fake news is the death of real news”Article URL: https://www.niemanlab.org/2026/07/now-were-getting-ai-fake-news-complaining-about-how-ai-fake-news-is-the-death-of-real-news/ Comments URL: https://news.ycombinator.com/item?id=48760598 Points: 153
Comments: 52
Section titled “Comments: 52”The gauge broke: devs felt 20% faster with AI, measured 19% slower (2025)
Section titled “The gauge broke: devs felt 20% faster with AI, measured 19% slower (2025)”Article URL: https://intrepidkarthi.com/writing/the-gauge-broke/ Comments URL: https://news.ycombinator.com/item?id=48757440 Points: 76
Comments: 97
Section titled “Comments: 97”2026-07-04
Section titled “2026-07-04”生成时间：2026/7/4 09:47:36（UTC: 2026-07-04T01:47:36.452Z）
Program-as-Weights: A Programming Paradigm for Fuzzy Functions
Section titled “Program-as-Weights: A Programming Paradigm for Fuzzy Functions”👍 58 · arXiv
Many everyday programming tasks resist clean rule-based implementation, such as alerting on important log lines, repairing malformed JSON, or ranking search results by intent, and are increasingly out…
AgenticSTS: A Bounded-Memory Testbed for Long-Horizon LLM Agents
Section titled “AgenticSTS: A Bounded-Memory Testbed for Long-Horizon LLM Agents”👍 41 · arXiv
Memory for a long-horizon LLM agent is a contract about what each future decision is allowed to see. The simplest contract appends past observations, tool calls, and reflections to every prompt, which…
EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments
Section titled “EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments”👍 40 · arXiv
Autonomous agents are increasingly expected to improve executable policies through feedback, yet existing evaluations often collapse this process into a final score or confound it with open-ended soft…
PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception
Section titled “PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception”👍 36 · arXiv
We introduce PerceptionRubrics, a rubric-based evaluation framework that addresses the gap between saturated benchmark scores and real-world brittleness. Shifting evaluation from holistic semantic mat…
Morphing into Hybrid Attention Models
Section titled “Morphing into Hybrid Attention Models”👍 31 · arXiv
Hybrid attention models improve long-context efficiency by retaining only a subset of full-attention layers and replacing the remaining layers with linear attention. However, the effectiveness of Tran…
OpenClaw v2026.7.1-beta.1
Section titled “OpenClaw v2026.7.1-beta.1”Highlights
Section titled “Highlights”- OpenAI GPT-5.6 support: OpenClaw now recognizes the GPT-5.6 model family across catalog, capability, and runtime selection paths. (#98333) Thanks @steipete-oai.
- **External har…
链接：https://github.com/openclaw/openclaw/releases/tag/v2026.7.1-beta.1
Goose v1.41.0
Section titled “Goose v1.41.0”✨ Features
Section titled “✨ Features”- iFlytek Spark and Astron MaaS providers (#9837)
--editsession flag to edit conversation before forking (#9799)- Fireworks AI declarative provider (#9990)
- OpenRouter request pa…
链接：https://github.com/aaif-goose/goose/releases/tag/v1.41.0
OpenAI Codex CLI rust-v0.143.0-alpha.35
Section titled “OpenAI Codex CLI rust-v0.143.0-alpha.35”Release 0.143.0-alpha.35
…
链接：https://github.com/openai/codex/releases/tag/rust-v0.143.0-alpha.35
Kagi Changelog (July 2): Heads, tails, and an AI toggle
Section titled “Kagi Changelog (July 2): Heads, tails, and an AI toggle”Article URL: https://kagi.com/changelog#10959 Comments URL: https://news.ycombinator.com/item?id=48779352 Points: 51
Comments: 10
Section titled “Comments: 10”AI saves about 3% of your hours, and almost none of it reaches the money
Section titled “AI saves about 3% of your hours, and almost none of it reaches the money”Article URL: https://okaneland.com/study/ai-productivity-roi-at-work/ Comments URL: https://news.ycombinator.com/item?id=48777257 Points: 70
Comments: 82
Section titled “Comments: 82”Instead of banning AI, I made a classroom contract with my students
Section titled “Instead of banning AI, I made a classroom contract with my students”Article URL: https://www.science.org/content/article/instead-banning-ai-i-made-classroom-contract-my-students Comments URL: https://news.ycombinator.com/item?id=48775499 Points: 69
Comments: 77
Section titled “Comments: 77”Please stop the AI confidence theater
Section titled “Please stop the AI confidence theater”Article URL: https://www.elenaverna.com/p/please-stop-the-ai-confidence-theater Comments URL: https://news.ycombinator.com/item?id=48774414 Points: 222
Comments: 240
Section titled “Comments: 240”AI Data Centers Use More Water Than Most Tech Giants Report
Section titled “AI Data Centers Use More Water Than Most Tech Giants Report”Article URL: https://www.wsj.com/tech/ai/ai-data-centers-water-use-901e2902 Comments URL: https://news.ycombinator.com/item?id=48774246 Points: 54
Comments: 64
Section titled “Comments: 64”Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says
Section titled “Alibaba to ban Claude Code in workplace over alleged backdoor risks, source says”Article URL: https://www.reuters.com/world/china/alibaba-ban-claude-code-workplace-over-alleged-backdoor-risks-source-says-2026-07-03/ Comments URL: https://news.ycombinator.com/item?id=48772443 Points: 314
Comments: 269
Section titled “Comments: 269”AI coding is a nightmare. Am I the only one experiencing this?
Section titled “AI coding is a nightmare. Am I the only one experiencing this?”Here are my biggest gripes with AI coding assistants right now: Obsessed with reinventing the wheel. You’ll often find it writing three duplicate functions for the exact same feature in a single file. Why? Because it’s terrified of blowing up the context window, so it only reads a fraction of a larg
Protect your right to run local AI
Section titled “Protect your right to run local AI”Article URL: https://righttointelligence.org/ Comments URL: https://news.ycombinator.com/item?id=48768951 Points: 496