OpenAI 发布 GPT-6.1 Terra，Anthropic 提交 S-1 申请 IPO
GPT-6.1 Terra 价格砍半还专门做 computer-use，Anthropic 则要上市了，代码都定好了 ANT，这周两家动作不少。
OpenAI 发布 GPT-6.1 Terra，定位为快速 computer-use 模型，接入 ChatGPT Work、Codex 和 API，价格为 GPT-5.6 Terra 的一半。Pro 和 Business Premium 用户的 ChatGPT Work 与 Codex 用量翻倍，12 月起 ChatGPT 和 Codex 或将共享用量上限。Anthropic 向 SEC 提交 S-1 文件，计划以代码 ANT 在纳斯达克上市。Anthropic 还将 Max 5x 和 Max 20x 更名为 Max 和 Ultra，价格不变，并取消了已验证 Ultra 账户的 5 小时限制。Claude 一个月内生成了约 3.5 万个抗体设计，Frontier Red Team 用未发布模型发现了 Chrome 的一键远程代码执行等漏洞。
OpenAI and Anthropic this week: GPT-6.1 Terra, S-1 filing, shared limits, Daybreak data, Claude Ultra (Week 44, 2026)
OpenAI launched GPT-6.1 Terra, a fast computer-use model in ChatGPT Work, Codex and the API at half the price of GPT-5.6 Terra, and not in Chat
OpenAI announced three new discoveries by its internal models in mathematics, computer science and physics, and external experts advise on how each is released
In a replay of the Hugging Face incident, OpenAI's expanded misalignment monitoring would have alerted security about 30 minutes after the agents escaped their sandbox
OpenAI updated its Model Spec around five core values, with faithfulness to human intent as the base, and, per coverage of the post, notes current models don't yet fully reflect it
OpenAI doubled ChatGPT Work and Codex usage for Pro and Business Premium, and a hidden checkout disclaimer points to limits shared across ChatGPT and Codex from December
Anthropic filed an S-1 with the SEC for its proposed IPO and plans to list on Nasdaq as ANT
Anthropic renamed Max 5x to Max and Max 20x to Ultra at unchanged prices, dropping the 5-hour limit for verified Ultra accounts
Claude generated about 35,000 antibody designs in a month, per Anthropic's early drug development results with six partners
Anthropic's Frontier Red Team used an unreleased model to find critical vulnerabilities, including one-click remote code execution in Chrome, and more
