Market · AI in practice
Anthropic gates Fable and Mythos 5.2 by verified purpose
On November 9, 2026, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, one model with two levels of access, priced 20% below the 5.1 generation and with outputs up to 30% faster. The physics results it cited are preprints that have not been peer reviewed.
What the sources report
On November 9, 2026, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, which share the same underlying model. Fable is available to all users, with strong cybersecurity and biology safeguards. Mythos goes to organizations approved through the Cyber and Life Sciences Verification Programs, which cover applications in cybersecurity and biomedical research. Pricing is $8 per million input tokens and $40 per million output tokens, against $10 and $50 for the 5.1 generation, according to the platform release notes. The company says outputs are up to 30% faster, and both models have a 1 million token context window.
From prerelease testing, Anthropic cites a formalization of Fermat's Last Theorem completed in 30 hours by about 1,000 Fable 5.2 agents and solutions to two open problems in black hole physics. On the official Claude account on X, the company adds that agents improved MIPLIB's best timetable for 42 trains by 2.6% and then proved no better timetable exists; the solutions were submitted to MIPLIB for checking. The physics results are published as preprints and have not yet gone through peer review. Anthropic itself recommends Claude Opus 5.5 as the starting point for most coding and knowledge tasks and keeps Fable and Mythos for the most ambitious technical work.
My take
In banking, access to a sensitive system is usually granted by role and by declared purpose, with an audit trail. Anthropic has carried that logic over to model capability: Fable reaches everyone with blocks on cybersecurity and biology, and anyone who needs the same model without those blocks goes through a verification program. I see a direct bridge in that design to risk and compliance teams, because the control sits at a point the institution already knows how to assess. With a lower price and an explicit access rule, the internal conversation about using frontier models in regulated sectors starts much further along.
Sources
- https://www.anthropic.com/claude-fable-and-mythos-5-2 (opens in a new tab) Anthropic
- https://platform.claude.com/docs/en/release-notes/overview (opens in a new tab) Claude Platform
- https://www.cnbc.com/2026/11/09/anthropic-claude-fable-5-2-mythos-5-2.html (opens in a new tab) CNBC
- https://x.com/claudeai/status/2119899482422911052 (opens in a new tab) Claude (X)