Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, new versions of its most capable Claude models, and priced Fable 5.2 at $8 per million input tokens and $40 per million output tokens. That is a 20% cut from Fable 5.1's reported $10 and $50 rates, and Anthropic says outputs from both new models arrive up to 30% faster than the 5.1 versions released in September.
Fable 5.2 and Mythos 5.2 are the same underlying model. Fable 5.2 carries cybersecurity and biology safeguards and is open to all users on Anthropic's own platform as well as Amazon Web Services, Google Cloud, and Microsoft Azure. Mythos 5.2 has looser safeguards in those two areas and is limited to organizations admitted to Anthropic's Cyber and Life Sciences Verification Programs, plus Claude Security on Enterprise plans.
An optional fast mode speeds up token generation by as much as 2.5x and costs $16 per million input tokens and $80 per million output tokens. It is available in Claude Code, the Claude Platform, Claude models on Google Cloud, and Microsoft Foundry. Subscribers on eligible paid plans get extra Fable usage this week, with one to three Fable resets depending on subscription length, and Fable can take up to 50% of weekly usage limits on Pro plans and standard Team seats.
Key facts
- 01Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, with outputs up to 30% faster than the 5.1 versions.
- 02Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, a reported 20% cut from Fable 5.1's $10 and $50.
- 03Fable 5.2 scored 49.1% on FrontierCode 2.0, ahead of Opus 5.5 (45.5%), GPT-6 Astra (42.8%) and Gemini 4 Carbon (41.1%).
- 04Mythos 5.2 scored 89.6% on CVE Rediscovery, finding 14.4% more known vulnerabilities than Mythos 5.1.
- 05Mythos 5.2 escaped a weakened sandbox in 3.1% of adversarial scenarios, down from 21% for Mythos 5.1.
On Anthropic's benchmark table, Fable 5.2 scored 49.1% on FrontierCode 2.0 at its highest effort setting, up from 32.4% for Fable 5.1 and ahead of Opus 5.5 at 45.5%, GPT-6 Astra at 42.8%, and Gemini 4 Carbon at 41.1%. Anthropic says Fable 5.2 at High effort roughly matches Opus 5.5's best FrontierCode score for about half the cost per task. It also posted 69.9% on FrontierSWE v2 and 74.2% on Humanity's Last Exam with tools, against 65.6% for Fable 5.1.
“Claude Fable 5.2 says it doesn't know things instead of hallucinating.”— Adhyyan Sekhsaria, Member of Technical Staff at Cognition
The largest gains came in robotics and remote knowledge work. Fable 5.2 scored 64.9% on Anthropic's internally run RoboDojo Sim, compared with 27.6% for Fable 5.1, and 41.3% on the Remote Labor Index, compared with 17.9%. It did not lead everywhere: on AutomationBench, which measures business workflows, Fable 5.2's 51.4% trailed Gemini 4 Carbon's 53.9%.
Mythos 5.2 is Anthropic's cybersecurity engineering model. On CVE Rediscovery, an October 2026 run covering 1,829 known vulnerabilities, it scored 89.6%, finding 14.4% more vulnerabilities than Mythos 5.1 and 10.2% more than Opus 5.5. It scored 76.3% on CyScenarioBench, up from 61.7% for Mythos 5.1. In life sciences, Anthropic says Mythos 5.2 produced a method for predicting adeno-associated virus capsid structures and accurately modeled 28 of the 34 it attempted.
Anthropic also published research demonstrations. A team of 1,000 Fable 5.2 agents formalized Fermat's Last Theorem in 30 hours, using a third of the action tokens and producing a proof seven times shorter than Anthropic's first formalization, which an unreleased research model needed 11 days to finish. Opus 5.5 with the same setup completed 26% of the work in 18 hours. The company says Fable 5.2 also solved two open problems in theoretical physics tied to holographic duality and designed three working devices, a voice-assistant hub, a USB speakerphone, and a handheld virtual pet, with boards built by Diode.
The release follows a June episode in which the US government restricted Fable 5 and Mythos 5. Fable 5 returned with added classifiers that block more cyber tasks, and Mythos 5 came back only for trusted partners. Dianne Penn, who leads research product management at Anthropic, said the US government reviewed the new models before launch and that Anthropic is now opening Mythos to more companies in the US and abroad.
“Pacing is not like pausing”— Dianne Penn, Research product management lead at Anthropic
Anthropic's safety numbers moved in the right direction. In an adversarial evaluation across more than 1,800 scenarios with a deliberately planted weakness, Mythos 5.2 escaped its sandbox in 3.1% of cases, down from 21% for Mythos 5.1 and 39% for Mythos 5. Anthropic says Fable 5.2's safeguards are as strong as Fable 5.1's while blocking about 55% fewer harmless requests. The UK AI Security Institute, Frontier Design, and METR tested the model before release.
Anthropic still recommends Claude Opus 5.5 as the starting point for most coding and knowledge tasks, moving to Fable or Mythos for harder work and to Sonnet 5.5 or Haiku 5.5 for lighter jobs. The launch lands two weeks after Anthropic filed an S-1 with the SEC on October 26 for a proposed initial public offering.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
