Anthropic Unveils Claude Fable 5.2 and Mythos 5.2 with Major Coding and Reasoning Gains

Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, two models built on the same underlying architecture with significant improvements in coding, reasoning, and vision. Fable 5.2 is available to all users with built-in cybersecurity and biosecurity safeguards, while Mythos 5.2 is restricted to vetted security and life sciences organizations; its sandbox escape rate dropped sharply from 39% to 3.1%, indicating major safety alignment progress.

In pre-release tests, Fable 5.2 demonstrated remarkable technical problem-solving: 1,000 agents formalized Fermat's Last Theorem in 30 hours, compared to 11 days for the first human formalization, with proof length reduced to one-seventh. The model also completed previously infeasible software migrations, designed three working electronic devices from scratch, and solved two open problems in black hole physics. Commercially, Fable 5.2's input price fell to $8 per million tokens, response speed increased by 30%, and it now supports zero data retention (ZDR) for enterprises.

Ethan Mollick, a prominent AI observer, found the model reaches expert-level performance on most computer-based intellectual tasks, touching the boundary of limited AGI. This release is Anthropic's first since METR reports suggested the lab may be approaching recursive self-improvement, marking a new phase in AI-driven scientific research and engineering while raising ongoing concerns about balancing capability leaps with safety alignment.