Anthropic's 1,000 Fable 5.2 Agents Formalize Fermat in 30 Hours

Anthropic on November 9 released Claude Fable 5.2 and Claude Mythos 5.2, pricing both at $8 per million input tokens and $40 per million output tokens, and calling them "best for the most ambitious, open-ended technical work."

The company says "a team of 1,000 Fable 5.2 agents fully formalized the theorem in just 30 hours, with minimal human intervention," referring to Fermat's Last Theorem, versus 11 days on an earlier version. Fable 5.2 also "solved two open problems in theoretical physics," both concerning black holes and leaning on holographic duality, "the idea that certain universes containing gravity can be precisely described by a quantum theory containing no gravity at all."

On security, Anthropic reports that "On our internal evaluation, CVE Rediscovery, Claude Mythos 5.2 found 10.2% more vulnerabilities than Claude Opus 5.5 and 14.4% more than Claude Mythos 5.1." The cost claim sits on a specific benchmark pair: "At High effort, Fable 5.2 scores about the same as Opus 5.5's best score (45.3% vs. 45.5%) for roughly half the cost per task."

Mythos is the same underlying model as Fable, offered with reduced safeguards to organizations enrolled in a Cyber Verification Program. The external endorsement comes through customer quotes rather than from any named Anthropic researcher: Mozilla distinguished engineer Brian Grinstead says "Mythos 5.2 roughly doubled the vulnerability discovery metrics of Mythos 5.1," and Vercel's Joe Haddad says "Projects we've been pushing along with other models for weeks got finished and approved by our teammates in a few days."

Anthropic also reports Fable 5.2 designed three working electronic devices and modeled 28 of 34 protein capsid structures during prerelease testing. The post names no Anthropic researcher as co-author of the Fermat or black-hole results. Across our Anthropic coverage the past three months, the company has shipped at a clip of several stories a week.