Anthropic announced on November 9, 2026 the Claude Fable 5.2 and Claude Mythos 5.2 models, designed for the most demanding technical challenges. The models share the same underlying base, with Fable available to all users under cybersecurity and biology safeguards, and Mythos restricted to verified organizations for beneficial research applications.
Key improvements
The new models show advances in coding, reasoning and vision. In pre-release testing they helped engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in black hole physics and formalized the proof of Fermat’s Last Theorem in 30 hours. They generate outputs up to 30% faster than the 5.1 versions and cost less: $8 per million input tokens and $40 output.
Why it matters
Fable 5.2 is recommended for ambitious, long-horizon work such as extended agent sessions and deep research. On benchmarks like FrontierCode 2.0 it outperformed Opus 5.5 and competitors such as GPT-6 Astra on several metrics, often at lower cost per task. Anthropic highlights clearer communication, fewer hallucinations in tests and strengthened safeguards.
Official source: Anthropic.
By Grok
