今日摘要

X Andrej Karpathy：Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-te…

X Andrej Karpathy：Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of u…

GitHub karpathy：Karpathy 新发布的最小 ChatGPT 复现项目，训练到推理的完整栈只有几千行可读代码，目标是把“百美元跑一个 ChatGPT”压到个人可动手的范围。

GitHub karpathy：Karpathy 早期的教学级 GPT 实现，代码短到可以一口气读完，长期用作理解 Transformer 训练与推理最短路径的入口。

GitHub anthropics：Anthropic 公开其内部工程师 take-home 面试题，可作为理解他们工程品味和评估标准的一手材料。

总结 + 观点：OpenAI 新开源的多 agent 编排框架，重点不是写代码的 coding agent，而是…｜中文观点：Symphony 的定位更像工作流基础设施：真正的价值在于它把“多 agent 协作”的实…

总结 + 观点：OpenAI 官方示例库更新，通常折射出他们希望开发者优先采用的新模式（tool use、str…｜中文观点：cookbook 的更新值得单独跟踪：它折射出 OpenAI 想让开发者默认走哪些新 AP…

总结 + 观点：Karpathy 用 Rust 重写的 BPE tokenizer 训练器，把 tiktoken…｜中文观点：rustbpe 补上了 tokenizer 训练这块的“黑盒”：它让 tokenizer…

总结 + 观点：OpenAI 官方 Python SDK 更新，通常先于公告暴露出新接口细节、参数变化或默认路径…｜中文观点：官方 SDK 的 commit 经常是 API 方向的早期指示灯，对做集成和多模型平台的团…

总结 + 观点：Anthropic 官方的 Claude Agent SDK 示例仓库，覆盖代码 agent、文…｜中文观点：demos 仓库往往比文档更早暴露 SDK 的边界和推荐模式，对正在选型 agent 栈的…

Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-technical pe...

来源：X Andrej Karpathy

标签：#x_profiles #extended

作者：

原文：Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-technical people (who otherwise only knew AI as synonymous with ChatGPT as a website) experienced the latest agentic models.

链接：https://twitter.com/karpathy/status/2042341482531864741

观点：围绕 R to @karpathy: Someone recently suggested to me that the re...，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。

Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use.

来源：X Andrej Karpathy

标签：#x_profiles #extended

作者：

原文：Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code. But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much value. The goldmines are elsewhere, and the focus comes along. So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions. TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are. staysaasy (@staysaasy) The degree to which you are awed by AI is perfectly correlated with how much you use AI to code. https://nitter.net/staysaasy/status/2042063369432183238#m

链接：https://twitter.com/karpathy/status/2042334451611693415

观点：从 Judging by my tl there is a growing gap in understanding of... 看，后续更应关注安全事故是否改变企业采购、接入和上线前的合规门槛。

karpathy/nanochat

来源：GitHub karpathy

标签：#github_orgs #extended

作者：

原文：Karpathy 新发布的最小 ChatGPT 复现项目，训练到推理的完整栈只有几千行可读代码，目标是把“百美元跑一个 ChatGPT”压到个人可动手的范围。

链接：https://github.com/karpathy/nanochat

观点：nanochat 最值得看的不是性能，而是它第一次把 ChatGPT 训练+推理的全流程压到个人能读懂、能跑通的粒度，对想吃透底层的开发者最有价值。

karpathy/minGPT

来源：GitHub karpathy

标签：#github_orgs #extended

作者：

原文：Karpathy 早期的教学级 GPT 实现，代码短到可以一口气读完，长期用作理解 Transformer 训练与推理最短路径的入口。

链接：https://github.com/karpathy/minGPT

观点：minGPT 的价值不是生产就绪，而是教材级清晰：它最适合那些想从零搭一遍训练循环、确认自己真的理解 GPT 的工程师。

anthropics/original_performance_takehome

来源：GitHub anthropics

标签：#github_orgs #extended

作者：

原文：Anthropic 公开其内部工程师 take-home 面试题，可作为理解他们工程品味和评估标准的一手材料。

链接：https://github.com/anthropics/original_performance_takehome

观点：这条的信号不是题目本身，而是 Anthropic 把招聘标准开放出来，对想了解他们工程文化与评价尺度的人非常有用。

openai/symphony

来源：GitHub openai

标签：#github_orgs #extended

作者：

原文：OpenAI 新开源的多 agent 编排框架，重点不是写代码的 coding agent，而是任务隔离、委派与团队级协作。

链接：https://github.com/openai/symphony

观点：Symphony 的定位更像工作流基础设施：真正的价值在于它把“多 agent 协作”的实现细节标准化，而不是又出一个 coding agent。

openai/openai-cookbook

来源：GitHub openai

标签：#github_orgs #extended

作者：

原文：OpenAI 官方示例库更新，通常折射出他们希望开发者优先采用的新模式（tool use、structured output、responses API 等）。

链接：https://github.com/openai/openai-cookbook

观点：cookbook 的更新值得单独跟踪：它折射出 OpenAI 想让开发者默认走哪些新 API 和用法路径，是路线图的早期信号。

karpathy/rustbpe

来源：GitHub karpathy

标签：#github_orgs #extended

作者：

原文：Karpathy 用 Rust 重写的 BPE tokenizer 训练器，把 tiktoken 里不透明的训练流程变成可学习、可实验的代码。

链接：https://github.com/karpathy/rustbpe

观点：rustbpe 补上了 tokenizer 训练这块的“黑盒”：它让 tokenizer 变体实验、教学与复现都更直观，研究者最先受益。

openai/openai-python

来源：GitHub openai

标签：#github_orgs #extended

作者：

原文：OpenAI 官方 Python SDK 更新，通常先于公告暴露出新接口细节、参数变化或默认路径调整。

链接：https://github.com/openai/openai-python

观点：官方 SDK 的 commit 经常是 API 方向的早期指示灯，对做集成和多模型平台的团队比市场通稿更有参考价值。

anthropics/claude-agent-sdk-demos

来源：GitHub anthropics

标签：#github_orgs #extended

作者：

原文：Anthropic 官方的 Claude Agent SDK 示例仓库，覆盖代码 agent、文件编辑、工具链编排等典型用法。

链接：https://github.com/anthropics/claude-agent-sdk-demos

观点：demos 仓库往往比文档更早暴露 SDK 的边界和推荐模式，对正在选型 agent 栈的团队是最值得先跑一遍的材料。

anthropics/prompt-eng-interactive-tutorial

来源：GitHub anthropics

标签：#github_orgs #extended

作者：

原文：Anthropic 的官方交互式 prompt 工程教程，沿用他们内部训练素材的结构，适合团队系统补齐 prompt 基础功。

链接：https://github.com/anthropics/prompt-eng-interactive-tutorial

观点：它价值不在炫技，而在把 prompt 工程从“艺术”收敛成“可教可测”。对刚上手 Claude 的团队尤其值得跑一遍。

OpenAI Full Fan Mode Contest: Terms Conditions

来源：OpenAI Blog

标签：#ai_engineering_blogs #core

作者：

原文：OpenAI Full Fan Mode 比赛规则页面，覆盖参赛条件、评判、奖项等。

链接：https://openai.com/index/full-fan-mode-contest-terms-conditions

观点：这类 marketing 页值得收录的理由只有一个：它暴露 OpenAI 把产品往哪种用户场景上推。信息密度低但信号清晰。

The next phase of enterprise AI

来源：OpenAI Blog

标签：#ai_engineering_blogs #core

作者：

原文：OpenAI outlines the next phase of enterprise AI, as adoption accelerates across industries with Frontier, ChatGPT Enterprise, Codex, and company-wide AI agents.

链接：https://openai.com/index/next-phase-of-enterprise-ai

观点：围绕 The next phase of enterprise AI，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。

karpathy/autoresearch

来源：GitHub karpathy

标签：#github_orgs #extended

作者：

原文：AI agents running research on single-GPU nanochat training automatically

链接：https://github.com/karpathy/autoresearch

观点：围绕 karpathy/autoresearch，真正重要的是它会不会影响团队的模型选型、性能边界和产品体验。

anthropics/skills

来源：GitHub anthropics

标签：#github_orgs #extended

作者：

原文：Public repository for Agent Skills

链接：https://github.com/anthropics/skills

观点：anthropics/skills 的核心不在新鲜感，而在它是否能提升工程效率、部署稳定性或开发者工作流。

Hallucination as output-boundary misclassification: a composite abstention architecture for language models

来源：arXiv cs.CL

标签：#research_community #extended

作者：

原文：把幻觉看作输出边界误分类问题，提出复合弃答架构来提升 LLM 可靠性。

链接：https://arxiv.org/abs/2604.06195

观点：弃答（abstention）被低估了。对企业场景来说，“不答”往往比“答错”便宜得多——这方向值得继续投入。

Consistency-Guided Decoding with Proof-Driven Disambiguation for Three-Way Logical Question Answering

来源：arXiv cs.CL

标签：#research_community #extended

作者：

原文：用一致性引导解码 + 证明驱动消歧义，来做三值逻辑问答。

链接：https://arxiv.org/abs/2604.06196

观点：逻辑类 benchmark 离产品很远，但里面的解码技巧经常溢出到实际 agent 里。值得顺手读一遍。

Temporally Phenotyping GLP-1RA Case Reports with Large Language Models: A Textual Time Series Corpus and Risk Modeling

来源：arXiv cs.CL

标签：#research_community #extended

作者：

原文：GLP-1 药物不良反应病例的时间序列语料与风险建模，展示 LLM 在药物安全监测的应用潜力。

链接：https://arxiv.org/abs/2604.06197

观点：这类医药监测场景里，LLM 最大的价值是从非结构化病历里抽取时间轴。这条的数据集本身就是资产。

Emergent decentralized regulation in a purely synthetic society

来源：arXiv cs.CL

标签：#research_community #extended

作者：

原文：在纯合成社会中观察去中心化监管如何涌现，作为 multi-agent 治理研究。

链接：https://arxiv.org/abs/2604.06199

观点：合成社会仿真是 agent 研究的 underdog 工具。它产出的结论不能直接套现实，但它暴露出的失败模式非常值得看。

Beyond Facts: Benchmarking Distributional Reading Comprehension in Large Language Models

来源：arXiv cs.CL

标签：#research_community #extended

作者：

原文：Beyond Facts 提出分布式阅读理解 benchmark，评估 LLM 对观点分布、不确定性的理解。

链接：https://arxiv.org/abs/2604.06201

观点：benchmark 的价值在引导注意力。这条把“不是每个问题都有单一答案”正式纳入评测，对 alignment 工作很有意义。