Anthropic has released Claude Fable 5.2, and it costs less than the model it replaces: $8 for every million tokens it reads and $40 for every million it writes. It is open to all users from 9 November, through Anthropic’s own platform and the three big clouds: AWS, Google Cloud and Azure.
The company launched a second model alongside it, Claude Mythos 5.2. Anthropic says the two are the same underlying model. Fable carries safeguards for cybersecurity and biology that let it go to all users, while Mythos is for organisations doing approved work in those areas.
The 5.1 pair arrived in September. Anthropic says the new ones do better on coding, reasoning and vision and write answers up to 30% faster. Buyers who want more speed can switch on fast mode, which Anthropic says generates tokens up to 2.5 times quicker. It doubles the rates to $16 for input and $80 for output per million tokens, and is offered in Claude Code, the Claude Platform, Google Cloud and Microsoft Foundry.
On its launch page, Anthropic puts Fable 5.2 at 49.1% on FrontierCode 2.0 at its highest effort setting, against 32.4% for Fable 5.1 and 45.5% for Claude Opus 5.5. It lists GPT-6 Astra at 42.8% and Gemini 4 Carbon at 41.1% on the same test. These are the company’s own figures, run with the safeguards switched on, and Anthropic’s chart does not give every rival the same effort level.
The page also describes what the company did with early versions. It says an engineer asked Fable 5.2 to build a prototype operating system from an empty folder, and the model coordinated more than 1,000 sub-agents over eight days to produce one that boots on a laptop. Anthropic adds that the models drew up three electronic gadgets that worked first time, each with a board ready to be made within a day.
The safeguards that make Fable public are unchanged in kind: classifiers refuse penetration testing, exploit writing and most biology research, and the request is answered by Claude Opus 4.8 (security) or Claude Opus 5 (biology) instead. What changes is how often they trigger. Anthropic’s early tests in its apps, Claude Code and Cowork saw roughly 55% fewer refusals than with Fable 5.1: about 60% fewer on security topics and 40% fewer on biology.
For everyday work, Anthropic itself points users to Claude Opus 5.5 first, and says to move up to Fable or Mythos only for the largest, most open-ended jobs, or down to Sonnet 5.5 or Haiku 5.5 for lighter ones. Fable 5.2 can also act as an adviser that smaller Claude models consult, or as an orchestrator that hands simpler tasks to them. Developers call it with the model name claude-fable-5-2.
Organisations that want Mythos 5.2 have to apply to Anthropic’s recently expanded Life Sciences and Cyber Verification programmes, and Mythos fast mode is reserved for members.
