面向开发者的产品、工具与工程实践更新。（本日 1 条）
Cloudflare 给 AI Gateway 加上实时预算阈值，重点看多模型调用怎么控成本 重大 topic|primary
Cloudflare Blog
GitHub 在 Copilot 多个体验里弃用 GPT-5.2 与 GPT-5.2-CodexGitHub Changelog / 2026-06-05这条更新的重点不是单纯下线两个模型名，而是 GitHub 开始重新整理 Copilot Chat、补全和 agent 模式背后的默认模型组合。真正值得看的，是哪些入口已经切走、哪些团队还需要补迁移、以及现有提示词、评测和自动化流程会不会被连带影响。
VS Code 开始公测企业统一管理插件能力GitHub Changelog / 2026-06-05VS Code 这次放出的不是一个新插件，而是企业统一管理插件装配和策略的入口，核心问题是开发环境治理终于开始产品化。对平台团队来说，更值得盯的是权限模型、允许名单、分发方式和团队级配置能不能真正落到现有 IDE 管理流程里。
Google 拆解 Gemini 企业 Agent 平台里的检索增强方案Google Research Blog / 2026-06-05
Google 这篇文章拆解的是企业检索增强的落地流程：先准备知识切片和索引，再处理查询改写、召回、重排与上下文拼装，目标是减少幻觉并稳住回答质量。更有用的是，它把效果验证和方法边界也放进同一套流程里，团队可以据此判断企业知识库问答到底该在哪些环节补检索和评测。
mvanhorn/last30days-skill 值得关注 NEW 今日 +1,111 stars: #1，较前一日持平 opencv/opencv 值得关注 NEW 今日 +65 stars: #2，较前一日持平 Leonxlnx/taste-skill 值得关注 NEW 今日 +1,103 stars: 用于约束 AI 输出风格、减少空泛生成的 skill 文件。；#3，较前一日持平 NousResearch/hermes-agent 一般 ↓ DOWN -2 今日 +1,112 stars: #4，较前一日下降 2 位 lfnovo/open-notebook 一般 ↑ UP +2 今日 +554 stars: #5，较前一日上升 2 位 yikart/AiToEarn 一般 NEW 今日 +183 stars: #6，较前一日持平 aaif-goose/goose 一般 NEW 今日 +322 stars: #7，较前一日持平 Crosstalk-Solutions/project-nomad 一般 NEW 今日 +309 stars: #8，较前一日持平
OpenRouter| Rank | Model | Provider | Value | Change |
|---|---|---|---|---|
| #1 | DeepSeek V4 Flash | deepseek | 3.69T tokens | 19% |
| #2 | Hy3 preview | tencent | 2.94T tokens | 3% |
| #3 | MiniMax M3 | minimax | 2.5T tokens | new |
| #4 | MiMo-V2.5 | xiaomi | 2.19T tokens | 50% |
| #5 | Owl Alpha | openrouter | 1.95T tokens | 17% |
| #6 | Claude Sonnet 4.6 | anthropic | 1.76T tokens | 9% |
| #7 | DeepSeek V4 Pro | deepseek | 1.7T tokens | 35% |
| #8 | Claude Opus 4.7 | anthropic | 1.44T tokens | 38% |
| #9 | DeepSeek V3.2 | deepseek | 1.17T tokens | 12% |
| #10 | Claude Opus 4.8 | anthropic | 1.16T tokens | 300% |
Artificial Analysis| 排名 | 模型 | 供应商 | 分数 | 指标 |
|---|---|---|---|---|
| #1 | Claude Opus4.8 (max) | anthropic | 61 分 | AA Index |
| #2 | GPT-5.5(xhigh) | openai | 60 分 | AA Index |
| #3 | GPT-5.5(high) | openai | 59 分 | AA Index |
| #4 | Claude Opus4.7 (max) | anthropic | 57 分 | AA Index |
| #5 | Gemini 3.1Pro Preview | 57 分 | AA Index | |
| #6 | Qwen3.7 Max | alibaba | 57 分 | AA Index |
| #7 | Gemini 3.5Flash | 55 分 | AA Index | |
| #8 | MiniMax-M3 | minimax | 55 分 | AA Index |
| #9 | Kimi K2.6 | moonshot | 54 分 | AA Index |
| #10 | MiMo-V2.5-Pro | xiaomi | 54 分 | AA Index |
榜首 Claude Opus4.8 (max)（anthropic），61 分，AA Index。
Guillermo Rauch@rauchgVercel AI Gateway recovers on average over 1T tokens a month 🤯 Much like Stripe recovers revenue with smart retries on failed payments or credit card updates. And we do it with 0️⃣ zero markup over the labs; adding r...
Simon Willison WeblogRelease: datasette-agent-edit 0.1a0 I'm planning several plugins for Datasette Agent which can make edits to existing pieces of text - things like collaborative Markdown editing, updating large SQL queries, and editin...
Simon Willison WeblogOpenAI Help: Lockdown Mode OpenAI first teased this in February , but now it's live and "rolling out to eligible personal accounts, including Free, Go, Plus, and Pro, and self-serve ChatGPT Business accounts": Lockdow...
Aaron Levie@levieThis is what the market got wrong about AI eating enterprise software. Building good software in the past was very hard. Yes, AI has made that a bit easier, though it’s still hard to build something that’s got good ta...
Aaron Levie@levieBox now has a markdown editor on the web. Full CLI support. Commenting. Full version history. Box Drive also lets you connect to any desktop client as a mounted drive, so you instantly work with all your files in Clau...
Thibault Sottiaux@thsottiauxI have a new kind of big button that I can press for Codex. Over the next 100 days, we will select one person per day who does impressive or incredibly useful work with Codex and give them 10X usage limits for a month...
