Anthropic 发布 Claude Fable 5.1 和 Mythos 5.1，称智能体任务最高便宜 45%

Anthropic 表示其最新的 AI 模型 Fable 5.1 和 Mythos 5.1：https://www.anthropic.com/claude-fable-and-mythos-5-1，解决了客户关于价格、数据保留和过度安全措施的批评。公司声称 Claude Fable 5.1：https://platform.claude.com/docs/en/models/fable-5-1/whats-new-fable-5-1 提供比 Fable 5 更强的性能，但通常价格低约 25%，在处理复杂代理任务时最多低 45%，这要归功于对已处理并存储的缓存数据的价格降低。

随着这一公告，一大批早期印象也随之出现，包括 Every CEO Dan Shipper，他声称：https://x.com/danshipper/status/2094848951568474186，“这是我们用过的最强的编码模型，但现在它速度快、令牌使用效率高，而且关键是实际上说话像正常人一样。”

Box CEO Aaron Levie 是另一个早期使用的拥护者：https://x.com/levie/status/2094851976769257770，说他公司使用 Fable 5.1 的代理在同一测试中捕捉到了 Fable 5 未能发现的数据中的微妙差异和模糊之处。与此同时，Lisan al Gaib 指出：https://x.com/scaling01/status/2094850290855936277，在基准测试中，低推理得分的 Mythos 5.1 与其前身的最大推理设置得分相同。

Anthropic 表示，它“现在允许 Fable 5.1 用于识别软件漏洞”，但仍会将一些网络安全任务重定向到 Opus 模型，如“渗透测试、漏洞利用生成和基于二进制的漏洞扫描”。Claude Fable 5.1 现已在所有平台上可用，而 Mythos 5.1 仅向 Project Glasswing 参与者开放。

Anthropic says its newest AI models, Fable 5.1 and Mythos 5.1：https://www.anthropic.com/claude-fable-and-mythos-5-1, address criticisms from customers about price, data retention, and overzealous safeguards. The company claims Claude Fable 5.1：https://platform.claude.com/docs/en/models/fable-5-1/whats-new-fable-5-1 offers stronger performance than Fable 5, but costs around 25 percent less typically and up to 45 percent less for complex agentic tasks, thanks to reduced pricing on cached data that was already processed and stored.

Along with the announcement, a slew of early impressions popped up, including from Every CEO Dan Shipper, who claims：https://x.com/danshipper/status/2094848951568474186, “It’s the strongest coding model we’ve used, but now it’s fast, token-efficient, and crucially actually speaks like a normal person.”

Box CEO Aaron Levie is another early access believer：https://x.com/levie/status/2094851976769257770, saying that his company’s agent with Fable 5.1 picked up on subtleties and ambiguities in data that Fable 5 missed in the same test. Meanwhile, Lisan al Gaib points out：https://x.com/scaling01/status/2094850290855936277 that on benchmarks, Mythos 5.1 with low reasoning scores the same as its predecessor set to Max reasoning.

Anthropic says it’s “now allowing Fable 5.1 to be used for identifying software vulnerabilities,” but it will still redirect some cybersecurity tasks to Opus models, like “penetration testing, exploit generation, and binary-based vulnerability scanning.” Claude Fable 5.1 is now available on all platforms, while Mythos 5.1 is available to Project Glasswing participants only.