Anthropic 11月9日发布 Claude Fable 5.2 与 Mythos 5.2：标准价每百万词元输入8美元、输出40美元，输出生成最高快30%
Anthropic 2026年11月9日发布新一代模型 Claude Fable 5.2 与 Claude Mythos 5.2。公司称，两者为同一底层模型的两种配置：Fable 5.2 面向一般用户开放，带有较强的网络安全与生物学防滥用防护；Mythos 5.2 放宽部分防护，只供通过审核的机构使用。
据 Anthropic 公布，Fable 5.2 即日起在 Claude 付费方案、Claude Code、Claude Platform 以及亚马逊云科技、谷歌云与微软 Azure 上线，开发者调用标识为 claude-fable-5-2。Mythos 5.2 的直接访问限于通过其网络安全审核计划与生命科学审核计划的机构；企业版客户另可经 Claude Security 使用 Mythos 做代码库漏洞扫描。
定价方面，两款模型的标准应用程序接口价格同为每百万输入词元8美元、每百万输出词元40美元，低于上一代 Fable 5.1 与 Mythos 5.1 的10美元与50美元。可选的快速模式价格为每百万输入词元16美元、每百万输出词元80美元，公司称其词元生成速度最高可达标准模式的2.5倍。
性能方面，公司称5.2版输出生成速度比5.1版最高快30%，沟通表达更清晰。平台文档显示，Fable 5.2 的上下文窗口为100万词元，最大输出12.8万词元，知识截止为2026年6月，支持文字与图像输入、文字输出。公司公布的 FrontierCode 2.0 评测中，Fable 5.2 在 Xhigh 档得分为49.1%，上一代为32.4%；在2026年10月的漏洞重现评测中，Mythos 5.2 对1,829个已公布漏洞的得分为89.6%，上一代为78.3%。上述分数均为公司公布的自测结果。
Anthropic 同时表示，Fable 5.2 在发布前测试中帮助处理了两个黑洞物理开放问题，相关预印本已经贴出；一组1,000个 Fable 5.2 智能体在30小时内以 Lean 形式化完成了费马大定理的证明形式化。公司并称，Fable 5.2 的防护强度与5.1版相当，早期测试中不必要的防护拦截比5.1版减少约55%，其中网络安全类减少约60%、生物学类减少约40%。公司仍建议多数用户日常编程与知识工作先从 Claude Opus 5.5 起步，Fable 与 Mythos 面向最具挑战性的开放式技术工作。路透社同日报道称，此次发布正值 Anthropic 筹备上市、与 OpenAI 竞争加剧之际。
Anthropic Launches Claude Fable 5.2 and Mythos 5.2 on November 9: Standard Pricing of $8 per Million Input Tokens and $40 per Million Output Tokens, With Output Generation Up to 30 Percent Faster
Anthropic released its new models Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026. The company said the two are configurations of the same underlying model: Fable 5.2 is the generally available version, carrying strong safeguards against misuse in cybersecurity and biology, while Mythos 5.2 relaxes some of those safeguards and is limited to vetted organizations.
According to Anthropic, Fable 5.2 is available in Claude paid plans, Claude Code, the Claude Platform, and on Amazon Web Services, Google Cloud and Microsoft Azure, under the developer model identifier claude-fable-5-2. Direct access to Mythos 5.2 is limited to organizations approved through Anthropic's Cyber Verification Program or Life Sciences Verification Program, and Claude Enterprise customers can also reach Mythos through Claude Security, which scans codebases for vulnerabilities.
Both models carry standard application programming interface pricing of $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1 and Mythos 5.1. An optional fast mode costs $16 per million input tokens and $80 per million output tokens and, the company said, can increase token generation speed by up to 2.5 times.
Anthropic said the 5.2 models generate output up to 30 percent faster than the 5.1 versions and communicate more clearly. Platform documentation lists a context window of 1 million tokens for Fable 5.2, a maximum output of 128,000 tokens, and a knowledge cutoff of June 2026, with text and image input and text output. In the company's published FrontierCode 2.0 evaluation, Fable 5.2 scored 49.1 percent at Xhigh effort, against 32.4 percent for Fable 5.1. In an October 2026 vulnerability rediscovery evaluation covering 1,829 published vulnerabilities, Mythos 5.2 scored 89.6 percent, against 78.3 percent for Mythos 5.1. The figures are the company's own published test results.
Anthropic also said that during prerelease testing Fable 5.2 helped address two open problems in black hole physics, with preprints posted for both results, and that a run of 1,000 Fable 5.2 agents formalized a proof of Fermat's Last Theorem in Lean in 30 hours. The company said Fable 5.2's safeguards are as robust as those in 5.1, with about 55 percent fewer unnecessary safeguard blocks in early testing, including about 60 percent fewer in cybersecurity and 40 percent fewer in biology. Anthropic continues to recommend Claude Opus 5.5 as the starting point for most everyday coding and knowledge work, with Fable and Mythos positioned for the most ambitious, open ended technical work. Reuters reported the same day that the launch comes as Anthropic prepares for a planned initial public offering amid intensifying competition with OpenAI.
来源链接
https://www.reuters.com/technology/anthropic-unveils-new-fable-mythos-models-competition-heats-up-ahead-ipo-2026-11-09/ （路透社2026年11月9日报道，数据与 Anthropic 当日发布口径一致）
https://www.securities.io/anthropic-launches-claude-fable-and-mythos-5-2-at-lower-api-prices/ （Anthropic 发布细节与平台文档数据汇总）
—————
来源：路透社
发布：GSM·美国（发布链接）
图片来源：Wikipedia/Wikimedia Commons 词条主图：Anthropic
GSM Global · 本篇英文版 English edition of this story
