Anthropic announced the release of its Claude Fable 5.2 and Claude Mythos 5.2 models on Monday, November 9, 2026, stating they are designed to address "daunting" technical challenges. The company reports that both models demonstrate advancements in coding, reasoning, and vision, and generate outputs up to 30% faster than the previous 5.1 versions. Pricing for the new models is set at $8 per million input tokens and $40 per million output tokens, a reduction from earlier versions. An optional "fast mode" is also available, which can increase token generation speed by up to 2.5 times at a higher cost.
Claude Fable 5.2 is available to all users across Anthropic's platforms, including the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure. Claude Mythos 5.2, which utilizes the same underlying model but with fewer safeguards, is restricted to verified organizations participating in Anthropic's Cyber Verification and Life Sciences Verification Programs for specialized cybersecurity and biomedical research applications.
During pre-release testing, Anthropic reported several notable achievements for the new models. These include solving two previously open theoretical physics problems related to black holes, designing three functional electronic devices from a single paragraph prompt, and formalizing the proof of Fermat's Last Theorem into a computer-verifiable form within 30 hours. The company also noted that a team of agents running an earlier version of Fable 5.2 successfully migrated a production service from Python to Rust, resulting in improved efficiency and reliability.
Anthropic positions Fable 5.2 and Mythos 5.2 for ambitious, open-ended technical work, such as agentic coding across large codebases, long-running agent sessions, and complex scientific research. For general coding and knowledge tasks, Anthropic recommends starting with its Claude Opus 5.5 model. On internal benchmarks, Fable 5.2 has surpassed Opus 5.5 in several evaluations, including computer use, knowledge work, cybersecurity, and agentic scientific research. In cybersecurity evaluations, Mythos 5.2 identified 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1.
The release of these new models follows a period of rapid AI model development by several companies, including OpenAI, Google, and Meta. Claude Fable 5.2 notably claimed the top spot on the Artificial Analysis Intelligence leaderboard on its launch day, a position previously held by Google Gemini 4 Carbon.