天呐，好突然，晓辉博士这次是真的下班了！😭 关注她有一段时间，很佩服她能每天下班路上边走边聊 AI 相关的内容，受益匪浅！ 以后再也听不到那句：“Hello 下班了”🥹祝福晓辉博士越来越好！

OpenAI 对 GPT-5.6 系列进行了发布以来最大的一次降价。 其中： GPT-5.6 Luna 输入价格下降 80% 成为了 OpenAI 目前最便宜的 GPT-5.6 模型 GPT-5.6 Terra 输入输出价格下降约 20% GPT-5.6 Sol价格保持不变 同时新增 High Speed API 模式，可以花更多钱换更低延迟。 Gemini 的性价比优势又被干翻 📊 性能没有缩水，反而继续提升 🏆 SWE-Bench Pro（代码） • Sol：64.6% • Terra：63.4% • Luna：62.7% • GPT-5.5：59.4% 🛡️ 网络安全（Capture the Flag） • Sol：96.7% • Terra：91.8% • Luna：85.2%

我现在用的最多的模型是： Claude Fable 5 GPT 5.6 Sol Claude Opus 4.6 Claude Opus 5 --- Opus 5 和 Sonnet 5 一样的毛病，过度思考，token 消耗很厉害，聪明不如 Fable 5，性价比不如 GPT 5.6 Sol，有点两头不沾，所以用的少，现在主要用在 Claude Design 相关任务。 Fable 5 虽然贵，但是结果相当靠谱，我会跟它一起讨论技术方案，或者复杂的任务。 其他的脏活累活则给 GPT 5.6 Sol，写作相关的还是 Opus 4.6 最好用。 昨天一个性能优化的任务，GPT 5.6 Sol 做了半天，最后给我悄悄把 16-bit 文本解码改成了 8-bit（图1），数字马上好看了，然后给我交差了😂 最后交给 Fable 5，虽然花了一些时间，但是找到问题根源（图2），还打脸了 GPT 5.6 的优化，最后优化好了。

🇨🇳This is bad news. Now, frontier open source access also will probably no more be available for all. China is preparing to limit foreign access to its strongest AI models, a move that could raise global AI costs and split the model market by nationality. Beijing has held recent talks with Alibaba, ByteDance, and Z .ai about keeping advanced Chinese models inside China, including models not yet released. The Ministry of Commerce led the discussions, with China’s state planning agency also present, which signals export control rather than routine platform regulation. The targets include closed models and open-weight systems, so the issue is not only API access but downloadable model power. Chinese officials also discussed treating leaks or theft of proprietary AI as a national security offence, not merely an IP dispute. New limits on who can fund Chinese AI startups were also discussed, which would tighten control over capital, talent, and model access together. Foreign companies could lose access to low-cost Chinese models just as those models become strong enough for serious production work. Washington has already restricted access to advanced U.S. models on security grounds. China now fears Mythos could find software vulnerabilities and be used against Chinese interests, so both sides are treating AI as strategic infrastructure. Beijing has also investigated Chinese AI startups that moved abroad and pushed Meta to unwind a $2B Manus deal. A likely path is tiered control: basic open tools get filings, stronger systems face reviews, and frontier models stay domestic. This would be a major setback for open AI access because model progress would no longer spread mainly through product quality and price. --- reuters .com/world/beijing-is-looking-curbing-overseas-access-chinas-top-ai-models-sources-say-2026-07-07/

Palantir 的 CEO 刚刚在CNBC的专访中控诉了 Sam Altman 和 Dario Amodei： “他们在抢劫每一家财富 500 强公司” 他的原话是： “这个国家里的每一家企业，这些人气得发疯。他们在为那些创造不出价值的代币付费。这些人正在窃取我业务的权重和核心优势。” 他直白地说，整个前沿 AI 商业模式就是披着订阅服务外衣的知识产权掠夺 然后他还用一个问题彻底摧毁了定价模型 “如果它这么有价值，假设我明天能让你赚 10 亿美元。我会不会说，我让你赚 10 亿美元，我要拿 30%？如果它这么有价值，为什么他们要按代币收费？” 如果 OpenAI 和 Anthropic 的模型真正实现了实验室声称的生产力提升，他们会选择股权或分成他们生成的利润。他们不会按百万token出售其服务 他把整个安排称为“一种不帮助穷人的财富税。它只是惩罚。” 美国企业正在将运营的核心优势——也就是工作流程、客户数据、战略备忘录、内部模型，这些让他们保持竞争力的东西——直接转移到少数硅谷实验室的训练管道中 一旦这些实验室重新训练，客户的独特优势就变成了下一个企业产品，反过来卖给他们的竞争对手 而 AI 行业最不想让人想到的部分： 每一家企业将机密文件、客户对话和财务模型输入前沿模型，都可能在教那个模型 HOW 如何取代他们。 供应商收取代币费用，同时积累关于该客户业务的复合智能。这就是机制。这也是为什么 Karp 用“stealing”这个词 他声称，这就是为什么他遇到的每位高管私下里都怒火中烧，公开却保持沉默。没人想成为那个点名实验室的 CEO，然后发现下一个竞争对手竟是用他们自己泄露的工作流程建成的 整个 AI 行业一直按完美定价，基于一个假设： 前沿实验室生产出持久、可防御的价值，足以证明无限计算支出的合理性 但 Karp 刚刚告诉我们，客户不再相信这个假设了。他们相信自己被征税却无益处，被监视却无同意，被复制却无救济。 企业一旦停止相信，整个估值体系就会动摇。

As we are still investigating, I have reset everyone's Codex usage limits. This is a hard reset given some users had stacked up to three banked resets already that they can apply on their own schedule. Funnily enough, this week at OpenAI is called the RESET week, which is meant for folks to relax a bit. However it will be a different kind of RESET week. Enjoy.

Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in supplemental training, is now in private beta at SpaceX & Tesla. Early evals show performance close to, perhaps exceeding Opus. RL is continuing to significantly improve the model, and the Grok Build harness gets better every day. Nice work by all those involved! Completely trained from scratch new models will be released by @SpaceX every month this year.