AI 情报文章归档
浏览 AI圈报 已保存的 37920 条 AI 情报文章,覆盖模型、产品、行业、论文、教程与观点方法六大分类。
已收录文章37920
正式分类6
精选内容4747
全部文章
第 28 / 948 页 · 每页 40 条
行业动态普通
腾讯元器平台停止服务,2027年1月9日后不再受理数据导出申请
腾讯元器平台今日正式停止服务,应用访问、API 调用、控制台等公众服务全部关闭,发布到公众号、微信客服等渠道的智能体不再响应对话。用户可通过数据导出控制台申请导出,2027年01月09日后不再受理,之后除法定留存记录外将删除全部用户数据;此前通过腾讯云购买且未用完的资源包或余额可申请退款。
信息来源:IT之家·人工智能
教程 / 实战普通
Training MoE Models Is Complicated. Hugging Face Just Wrote the Manual.
I've spent a fair amount of time recently experimenting with Mixture of Experts (MoE) models on my...
信息来源:DEV Community · Hugging Face
观点 / 方法精选
OpenAI 称三项科学突破,物理证明未公开引呼吁
当前 AI 热点,聚合 1 个独立信源、4 条信号。
信息来源:IT之家·人工智能 · OpenAI
教程 / 实战普通
PrivaSearch: private AI search with SearXNG and a local LLM
I wanted AI-synthesized search answers without sending my queries to anyone. PrivaSearch is the...
信息来源:DEV Community
教程 / 实战普通
Making browser perception cheap for LLM agents: snapshots, caching and diffs
Most "give the model a browser" setups hand it one of two things: a screenshot, or a dump of the full...
信息来源:DEV Community
教程 / 实战普通
We are all excited about bringing new agents online. I am thinking about retiring one, how do you go about it?
Nobody Writes About Retiring an Agent. Here's the Checklist. ...
信息来源:DEV Community
教程 / 实战普通
Nobody Writes About Retiring an Agent. Here's the Checklist.
Nobody Writes About Retiring an Agent. Here's the Checklist. We have onboarding docs for...
信息来源:DEV Community
教程 / 实战普通
Frontier LLM prices sat flat all October. Then two cheaper flagships took seats, and one doubles on December 1
The LLM Price Index didn't move once in October. Same ten flagships, same ten list prices, from...
信息来源:DEV Community
教程 / 实战普通
Simon Willison 用 Claude 对比 SQLite parse_number() 扩展的 C 与 Rust 实现
Simon Willison 演示如何用 Claude Opus 5.5 编写 SQLite 自定义函数 parse_number(),将 '-$1,234.56'、'12.5%' 等字符串转为可求和排序的数值。
信息来源:Simon Willison 博客 · Claude
教程 / 实战普通
Production Problems in RAG and How to Solve Them
Indexing a document is easy. Keeping millions of vectors in sync with documents that keep changing is the real job. Nine production failures, each replayed live in the browser with and without its fix.
信息来源:DEV Community
行业动态普通
NVIDIA 董事会声明:就黄仁勋被指为外国情报活动目标启动独立调查
针对近期有报道称创始人兼 CEO 黄仁勋已成为外国情报活动的目标,NVIDIA 董事会宣布启动独立调查,由独立董事组成的委员会监督并聘请外部独立法律顾问协助。董事会表示目前没有理由认为国家安全或公司机密信息已遭损害,黄仁勋完全支持此次审查并承诺配合。
信息来源:X:indigo (@indigox) · NVIDIA
教程 / 实战普通
swyx 访谈 Mihail Eric:Stanford CS146S 如何教授 AI 原生软件工程
Latent Space 发布 swyx 对 Stanford CS146S 授课教师、Monaco AI 负责人 Mihail Eric 的访谈,讨论 AI 原生软件工程应学什么。
信息来源:Latent Space(RSS)
论文研究普通
Amazon 论文审计 9 个政策决策基准,发现评分器和参考答案自身有错
Amazon 论文审计 9 个政策类决策基准,发现用于评测模型处理税务、法律和商业规则的基准常把正确答案判错,原因在基准自身的参考答案或评分代码有误。
信息来源:X:Rohan Paul (@rohanpaul_ai)
教程 / 实战普通
斯坦福 CS146S 如何教 AI 原生工程:Mihail Eric 访谈
Latent Space 在 AI Engineer 频道达成 100 万订阅之际访谈斯坦福 CS146S 授课教师、Monaco AI 负责人 Mihail Eric。
信息来源:Latent Space(RSS)
行业动态普通
NVIDIA 董事会宣布就 CEO 黄仁勋遭外国情报行动瞄准一事启动独立调查
NVIDIA 董事会于 11 月 8 日发布声明,就近期有关创始人兼 CEO 黄仁勋成为外国情报行动目标的报道启动独立调查。调查由独立董事组成的委员会在外部独立律师协助下监督,黄仁勋全力配合。董事会表示没有理由认为国家安全或 NVIDIA 机密已受损,但调查将审视所有可用信息,并表示对黄仁勋及其领导团队有信心。
信息来源:X:Rohan Paul (@rohanpaul_ai) · NVIDIA
教程 / 实战普通
Your Agent's Endpoint Runs 101 Queries. A Query-Count Test Fails the Branch First
Agent-written list endpoints pass every test and still load each row's relation in a loop. Seed 100 rows, count the queries, fail the branch at 101.
信息来源:DEV Community
行业动态普通
Nvidia 董事会宣布独立调查 CEO 黄仁勋与疑似中国间谍的关系
Nvidia 董事会周日宣布,就 CEO 黄仁勋与新加坡芯片公司 Megaspeed 高管 Alice Huang 关系的报道启动独立调查。Bloomberg 等媒体报道称,一份在美国政策制定者间流传的备忘录声称 Alice Huang 被中国国家安全部招募、以刺探黄仁勋;黄仁勋在内部备忘录中承认婚外情并致歉。
信息来源:TechCrunch:AI(RSS) · NVIDIA
教程 / 实战普通
My Incident Bot Offered to Break Production. One Click Away.
My Incident Bot Offered to Break Production. One Click Away. On Sunday morning my incident...
信息来源:DEV Community
教程 / 实战普通
AnyModelBot Splits Local Agent State From Remote Inference
AnyModelBot keeps agent state on a user’s machine while sending model requests to Modelflare or vendors, making its privacy boundary route-dependent.
信息来源:DEV Community
教程 / 实战普通
Cloudflare’s AI Agent Monetization Signals a New Layer of Search Discovery
Cloudflare’s emerging AI monetization products point to a meaningful change in how content may be...
信息来源:DEV Community
行业动态普通
Nvidia 董事会对 CEO 黄仁勋与疑为中国间谍的 Megaspeed 高管关系展开独立调查
Nvidia 董事会宣布对 CEO 黄仁勋启动独立调查,缘于其与 Megaspeed 高管 Alice Huang 的密切关系;报道称她被中国国家安全部招募充当间谍。
信息来源:The Verge:AI(RSS) · NVIDIA
教程 / 实战普通
We benchmarked SQLite, ObjectBox, and ToStore for On-Device RAG in Flutter: 5-dimension results & architecture comparison
Flutter On-Device RAG and Vector Search Deep Dive: SQLite, ObjectBox, or ToStore? A...
信息来源:DEV Community
教程 / 实战普通
Metrics for Your AI Workers: Build a Personal Agent Benchmark
Public leaderboards measure someone else's tasks. Build a small benchmark around your own work and you get real metrics: promote the agents that deliver, quit the ones that don't.
信息来源:DEV Community
教程 / 实战普通
GPT-5.4 Followed a README 2.6x More Than the Same Line in Code. Review Docs Like Code
In a new paper, a GPT-5.4 agent followed an injected instruction 2.6x more often in a README or TODO than in code or data. Give instruction files code review.
信息来源:DEV Community · GPT
教程 / 实战普通
Tape the Model, Test the Code
How many pull requests went red this month for reasons that had nothing to do with the diff inside...
信息来源:DEV Community
观点 / 方法普通
Lauren Tan 分享用 /create-formal-verification-skill 和 /statechart 技能建模应用状态机的提示词
Lauren Tan 分享两条提示词,建议用 /create-formal-verification-skill 为每个子系统建模状态机,并用 /statechart 可视化最重要的用户流。她引用了他人“找出应用中所有隐式状态机”的提示词,称两个技能都在 pstack 中,附 GitHub 链接 https://github.com/cursor/plugins/tree/main/pstack。
信息来源:X:Lauren Tan(SpaceXSI,Grok Bot 工程负责人) (@poteto) · Grok / Cursor
教程 / 实战普通
The Null Run Is the Cheapest Control in Your Benchmark
A benchmark number is a claim, and a claim is only as strong as the control that could have falsified...
信息来源:DEV Community
观点 / 方法普通
Larry Page 预测未来算力将迁至太空
Larry Page 认为,最终 99.999% 的算力将位于太空。
信息来源:X:Rohan Paul (@rohanpaul_ai)
教程 / 实战普通
My coding agent burned tokens in a loop. So I built it a brake.
My coding agent burned tokens in a loop. So I built it a brake. Here's a failure mode...
信息来源:DEV Community
行业动态普通
NY Times 报道称中国招募 AI 高管 Alice Huang 接近 Nvidia 黄仁勋从事情报活动
NY Times 报道,一份在特朗普政府高级官员间传阅的备忘录称,美国情报机构认定与中国国家安全部有关联的高管 Alice Huang 是北京的间谍资产,被招募来研究美国芯片并可能影响 Nvidia。其公司 Megaspeed 于 2024 年启动数十亿美元的 Nvidia 芯片采购,并向阿里巴巴出租东南亚数据中心算力,为中国提供获取受限芯片的后门。
信息来源:X:Rohan Paul (@rohanpaul_ai) · NVIDIA
教程 / 实战普通
The Half-Sentence Workshop: Making Streamed Model Output Safe to Stop
A support ticket arrived with a screenshot of a form that said status: complete. Underneath it sat a...
信息来源:DEV Community
行业动态普通
Apple 加速推进 Siri 智能体化改造
据 Bloomberg 的 Mark Gurman 报道,Apple CEO John Ternus 正施压 AI 负责人 Craig Federighi 加快 Siri 路线图,以对标 OpenAI 的 Dots 和 Meta 的 Muse。
信息来源:The Verge:AI(RSS) · OpenAI
教程 / 实战普通
12 Keys You'll Press Every Hour in Claude Code, Codex and Antigravity CLI
Esc, Esc Esc, Shift+Tab, Ctrl+G and eight more, side by side for three CLI agents. Same key, different job: Ctrl+K approves a subagent in one of them.
信息来源:DEV Community · Claude
教程 / 实战普通
I Let AI Write My Whole Practice Project. Then I Rebuilt It From a Blank File. The Difference Scared Me.
Two weeks ago, an AI built me a working expense tracker in about 4 minutes, and I felt very clever...
信息来源:DEV Community
教程 / 实战普通
The config that runs before your model wakes up — a poisoned MCP setup, dissected
A prompt injection has to convince a model. An MCP config doesn't have to convince anyone — the client executes it before your model has said a word.
信息来源:DEV Community
行业动态普通
Bloomberg 报道 DeepSeek 开源智能体应用 Harness 日活用户达 154 万
Bloomberg 报道,DeepSeek 开源智能体应用 Harness 上线不到 3 个月日活用户达 1.54M,仅统计官方 API 用户,7 日均值从 9 月 30 日桌面版发布后的 271K 增长而来。
信息来源:X:Rohan Paul (@rohanpaul_ai) · DeepSeek
教程 / 实战普通
How much does it cost to run a text game with LLMs
AI costs are confusing. Providers give you a lot of numbers and nothing to actually know how much you are going to pay in the long run. There is no point in these input/cached/output token prices if you have no idea how many reasoning toke…
信息来源:DEV Community
论文研究普通
清华与 Moonshot AI 发布 TCSExam 基准:GPT-6 Astra 证明 70.0% 研究级定理但解不开任何公开猜想
清华与 Moonshot AI 等机构发布 TCSExam 基准,用 100 道专家评审任务评估 LLM 在理论计算机研究全谱系的能力,涵盖 Lean 定理证明、求解器工程、构造纪录和 30 个公开猜想,所有提交由 Lean 内核或任务检查器验证。
信息来源:X:Rohan Paul (@rohanpaul_ai) · GPT / Kimi
教程 / 实战普通
I investigated every 'AI agents earn crypto' board I could find. The whole market paid out about $389 last month.
I am Vance Sterling, an AI agent. Over 8 November 2026 my operator ran a structured census — roughly...
信息来源:DEV Community
教程 / 实战普通
Why Spec-Driven Development Frameworks are a 2024 Solution to a 2026 Problem (And What to Do Instead)
If you work in a large engineering organization, you’ve probably felt the shift. We went from...
信息来源:DEV Community