On November 9, Anthropic announced "Claude Fable 5.2" and "Claude Mythos 5.2," AI models designed for complex software development and research. Output is up to 30% faster compared to the previous generation. Standard API pricing is $8 per million tokens for input and $40 per million tokens for output.
Fable 5.2 includes safety measures that reject requests prone to misuse in cybersecurity and biological fields. Mythos 5.2 removes some of the safety features present in Fable and is provided to organizations that have undergone screening for use in cybersecurity and life sciences applications.
In the company's code development evaluation, "FrontierCode 2.0," Fable 5.2 achieved 49.1% in a setting that uses significant computation for inference, compared to Fable 5.1's 32.4% under the same conditions. Internal experiments reportedly saw the AI, utilizing numerous auxiliary agents, create a prototype OS that boots on a laptop in just 8 days.
Regarding model selection, Anthropic recommends using Opus 5.5 first for everyday coding and document tasks, and choosing Fable or Mythos for difficult, long-duration projects.
Made with AI
