Claude Fable 5.2 and Claude Mythos 5.2
Anthropic released Claude Fable 5.2 and Mythos 5.2, and says a team of 1,000 Fable 5.2 agents formalized Fermat's Last Theorem in 30 hours.
Both models are upgrades of Fable and Mythos, which Anthropic describes as best for the most ambitious, open-ended technical work, with outputs up to 30% faster than 5.1. Anthropic says that in prerelease testing the models formalized the proof of Fermat's Last Theorem and solved two open black hole physics problems. In its Fermat test, Opus 5.5 with the same setup completed only 26% in 18 hours.
- Date
- Monday 9 November 2026
- Lab
- Anthropic
- Kind
- model
- Access
- closed API
Figures
| Measure | Value | Measured by |
|---|---|---|
| Output speed versus 5.1 | up to 30% faster as stated in the Techmeme summary of Anthropic's announcement | company |
| Fermat's Last Theorem formalization | 1,000 Fable 5.2 agents, 30 hours Opus 5.5 with the same setup reached 26% in 18 hours | company |
| Capsid structures modeled accurately by Mythos 5.2 | 28 of 34 Anthropic's own tests | company |
VentureBeat's headline, as listed on Techmeme, mentions a 20% API cost reduction. Not confirmed on a primary page. This is separate from the September 2026 Claude agents Fermat Lean formalization record.
Sources
This record was partly confirmed: some claims could not be checked on 9 November 2026. How we check