Anthropic ships Claude Fable 5.2 and vetted-tier Mythos 5.2

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, at $8 per million input tokens and $40 per million output tokens, with an optional fast mode at double those rates. The two names share an underlying model; what differs is who can run it.

"Claude Fable and Claude Mythos are the same underlying model," Anthropic writes in the announcement. "Fable has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users. Organizations pursuing beneficial applications in cybersecurity or biomedical research can apply for access to Mythos through our recently expanded Life Sciences and Cyber Verification Programs." Four researchers in our Who's Who tracker had shared the post by the time it reached our feed.

On benchmarks, Fable 5.2 at Max effort scores 64.9% on the RoboDojo robotics simulator, which Anthropic calls "more than double GPT-6 Astra (29.8%) and Fable 5.1 (27.6%)." On CVE Rediscovery the company reports Mythos 5.2 "found 10.2% more vulnerabilities than Claude Opus 5.5 and 14.4% more than Claude Mythos 5.1." Outputs are "up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1."

On safeguards the dial moves the other way. The company says it "observed about 55% fewer blocks with Fable 5.2 compared with Fable 5.1."

Anthropic leans on agent-swarm demos to make the capability case. "a team of 1,000 Fable 5.2 agents fully formalized the theorem in just 30 hours, with minimal human intervention," the announcement says of Fermat's Last Theorem. For contrast: "Our first formalization of Fermat's Last Theorem, performed by an unreleased research version of Claude, took 11 days." A separate demo migrated a production service from Python to Rust; Anthropic says it now "starts about 26 times faster."

"Claude Mythos 5.2 is astonishingly good," says Brian Grinstead, a Distinguished Engineer at Mozilla, in a customer testimonial published alongside the release.