模型、公司、政策与基础设施层面的行业动向。（本日 4 条）
OpenAI 模型与 Codex 可通过 Oracle 云承诺访问 重大 topic|primary coding agent: 7d 26x/7d
OpenAI News RSS OpenAI 报告关联中国的影响行动瞄准美国 AI 讨论 重大 topic|primary
OpenAI News RSS
DeepMind 推出统一的无编码器多模态 Gemma 4 12B 重大 topic|primary
Google DeepMind RSS
Google 用新 Gemini 工具面向企业节省时间并促进增长 重大 topic|primary
Google Keyword Blog面向开发者的产品、工具与工程实践更新。（本日 3 条）
Meta 解释 AI 基础设施中的算力角色 重大 topic|primary
Meta Newsroom
Infrastructure Explained: Compute Power：图片来自 Meta Newsroom，由 feed 提供，用于辅助理解该条目的产品、研究或内容生成语境。
阿里云公开 Agent 可观测与审计数据采集实践 重大 topic|primary
Alibaba Cloud Blog OpenAI 展示 Codex 辅助天体物理学家模拟黑洞 重大 topic|primary coding agent: 7d 26x/7d
OpenAI News RSS值得关注的开源模型与项目动态。（本日 1 条）
Google：年轻人应参与塑造 AI 的未来 重大 topic|primary
Google Keyword Blog
GitHub Copilot Chat 已可查看 agent 会话GitHub Changelog / 2026-06-10文章拆解 agent、开发工具或自动化流程里的任务规划、权限、上下文、工具调用和失败恢复。原文把价值落在代码、接口、README、案例和失败模式这些硬信息上，而不是只给观点。这类文章最有用的地方，是能帮团队判断它该不该进入试点、采购或内部自动化路线图。
英伟达说明如何高吞吐运行 DiffusionGemmaNVIDIA Developer Blog / 2026-06-10文章拆解 agent、开发工具或自动化流程里的任务规划、权限、上下文、工具调用和失败恢复。真正的信息密度在真实场景、接入门槛、价格、可用地区、案例证据和工作流限制。这类文章最有用的地方，是能帮团队判断它该不该进入试点、采购或内部自动化路线图。
GitHub 让 Copilot CLI 借助语言服务器理解真实代码GitHub Blog Feed / 2026-06-10文章拆解 agent、开发工具或自动化流程里的任务规划、权限、上下文、工具调用和失败恢复。原文把价值落在代码、接口、README、案例和失败模式这些硬信息上，而不是只给观点。它能帮助读者判断这条变化是否会影响产品路线、工具选型或内部风险预案。
addyosmani/agent-skills 值得关注 NEW 今日 +821 stars: #1，较前一日持平 phuryn/pm-skills 值得关注 ↑ UP +6 今日 +804 stars: #2，较前一日上升 6 位 refactoringhq/tolaria 值得关注 ↑ UP +2 今日 +612 stars: #3，较前一日上升 2 位 mvanhorn/last30days-skill 一般 ↓ DOWN -3 今日 +2,535 stars: #4，较前一日下降 3 位 soxoj/maigret 一般 NEW 今日 +318 stars: #5，较前一日持平 x1xhlol/system-prompts-and-models-of-ai-tools 一般 ↑ UP +3 今日 +393 stars: #6，较前一日上升 3 位 obra/superpowers 一般 NEW 今日 +1,104 stars: #7，较前一日持平 masterking32/MasterDnsVPN 一般 NEW 今日 +354 stars: #8，较前一日持平
OpenRouter| Rank | Model | Provider | Value | Change |
|---|---|---|---|---|
| #1 | DeepSeek V4 Flash | deepseek | 4.34T tokens | 55% |
| #2 | Hy3 preview | tencent | 3.79T tokens | 36% |
| #3 | MiniMax M3 | minimax | 3.38T tokens | 343% |
| #4 | MiMo-V2.5 | xiaomi | 2.89T tokens | 27% |
| #5 | DeepSeek V4 Pro | deepseek | 2.06T tokens | 58% |
| #6 | Claude Sonnet 4.6 | anthropic | 2.03T tokens | 13% |
| #7 | Owl Alpha | openrouter | 2.01T tokens | 4% |
| #8 | Claude Opus 4.7 | anthropic | 1.71T tokens | 13% |
| #9 | Claude Opus 4.8 | anthropic | 1.32T tokens | 77% |
| #10 | DeepSeek V3.2 | deepseek | 1.18T tokens | 7% |
Artificial Analysis| 排名 | 模型 | 供应商 | 分数 | 指标 |
|---|---|---|---|---|
| #1 | Claude Fable 5 (with fallback) | anthropic | 65 分 | AA Index |
| #2 | Claude Opus 4.8 (max) | anthropic | 61 分 | AA Index |
| #3 | GPT-5.5 (xhigh) | openai | 60 分 | AA Index |
| #4 | GPT-5.5 (high) | openai | 59 分 | AA Index |
| #5 | Claude Opus 4.7 (max) | anthropic | 57 分 | AA Index |
| #6 | Gemini 3.1 Pro Preview | 57 分 | AA Index | |
| #7 | Qwen3.7 Max | alibaba | 57 分 | AA Index |
| #8 | Gemini 3.5 Flash | 55 分 | AA Index | |
| #9 | MiniMax-M3 | minimax | 55 分 | AA Index |
| #10 | Kimi K2.6 | moonshot | 54 分 | AA Index |
榜首 Claude Fable 5 (with fallback)（anthropic），65 分，AA Index。
英伟达为本地 AI 加速 DeepMind 的 DiffusionGemmaNVIDIA Newsroom RSS / 2026-06-10
文章梳理一个 AI 产品、平台或工程实践的具体变化，而不是只给观点。真正的信息密度在真实场景、接入门槛、价格、可用地区、案例证据和工作流限制。它能帮助读者更新对模型能力的预期，而不是只记住排行榜或单个指标。
Swyx@swyxfor those keeping track at home it was 34 days between signing this deal and launching Mythos-class model GA to the world. https://t.co/J1dZDZcxMu building on @nvidia stack means you can just do things™.
Boris Cherny@bchernyFable 5 is the biggest step up I’ve felt in our models since Opus 4.5 back in November. After 4.5 came out I uninstalled my IDE when I realized that I’d been doing 100% of my coding in a terminal for a few weeks. With...
Guillermo Rauch@rauchgVercel CLI now allows you to: ◾ create AI Gateway API keys ◾ pass a --𝚋𝚞𝚍𝚐𝚎𝚝 to cap their spend ◾ set a --𝚛𝚎𝚏𝚛𝚎𝚜𝚑-𝚙𝚎𝚛𝚒𝚘𝚍 for the quota Think of it as virtual credit cards for AI tokens 🤖💳 https://...
Andrej Karpathy@karpathyThis is a super exciting release - Claude Fable 5 is the same underlying model as Mythos but with added safeguards. The benchmarks are great and it's SOTA on everything by a margin but I'll add that qualitatively al...
Swyx@swyxbtw insane amounts of alpha in telling claude code to "review my code for issues" on Fable rn while it is not pay per use be prepared to be in abject horror that you shipped anything to prod without a Fable Check™ fir...
Boris Cherny@bchernyWe talk a lot about how important it is to set up self-verification loops. Especially in the age of powerful models that can run for long periods of time, self-verification is a key ingredient that enables the model t...
Thibault Sottiaux@thsottiauxDo you use codex /goal occasionally or as your main way to get things done?
Thibault Sottiaux@thsottiauxPlaying codex like an orchestra. One /goal at a time. https://t.co/vlpypZu20A
Peter Yang@petergyangwtf does "big model smell" mean
Peter Yang@petergyangBtw this is the prompt I used: Build F-Zero: futuristic anti-gravity racer with pseudo-3D track (raycasting or mode-7 style scaling), 3 AI opponents, boost meter that drains health, speed 400-800 km/h with visible sen...
Thariq@trq212at Code w/ Claude Tokyo! say hi if you see me around https://t.co/BXfvEJVvrf
Thariq@trq212Fable is a step-change in models, and I hope it changes how you work with Claude. More to come in a series of posts on how it’s reshaped our work, but the TLDR: it’s time to be more ambitious. https://t.co/bWoxbTBShh
