Claude Mythos 5.1: fewer refusals for vetted security and bio work
It's the same model as the new Fable 5.1 — same price, same benchmarks — but a life-sciences track now joins cyber vetting to loosen Claude's guardrails.
Copy markdown
Same model as Fable 5.1, just fewer refusals
Anthropic says Mythos 5.1 is identical to the newly shipped Fable 5.1 — same $10/$50 per Mtok, same 75%-cheaper cache reads at $0.25/M — with one difference: “more permissive safeguards for vetted individuals and organizations” whose legitimate work keeps tripping Claude's guardrails mid-task.
The new door is life sciences, not just cyber
Access runs through two vetting tracks: the Cyber Verification Program for defensive-security work, and a Life Sciences Verification Program for bio and biotech pros — US organizations only for now. If your legitimate vuln research or life-sciences modeling keeps hitting refusals, this is the sanctioned path to the same model with fewer of them.
Everyone else already has the capability
You're not missing the model — only the permissive tier. Fable 5.1 (`claude-fable-5-1`) is generally available on the Claude API, Bedrock, Vertex, Azure Foundry, Claude.ai and Claude Code, and posts the same scores: 55.8% on Terminal-Bench 4.0, 73.4% on CursorBench, and 52.6% on Terminal-Bench-Science — more than double Fable 5's 24.7%.