Anthropic ships Claude Fable 5.2 to all customers and restricts the same model, as Mythos 5.2, to verified cyber and life-science users

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9 as its models for difficult coding, finance and science work. The company's platform documentation lists the same 1-million-token context window and the same $8 and $40 per-million-token pricing for both. Coverage describes them as one underlying model with two access regimes. Fable 5.2 is generally available with safeguards. Mythos 5.2 is limited to organizations that pass Anthropic's cyber or life-sciences verification programs.

Anthropic says both models generate output up to 30% faster than the 5.1 versions. In an internal evaluation, Mythos 5.2 found 14.4% more known vulnerabilities than Mythos 5.1. Anthropic's 14-benchmark table shows Fable 5.2 leading on most tests against Fable 5.1, Opus 5.5, GPT-6 Astra and Gemini 4 Carbon. On Terminal-Bench-Science 0.1 it reports 74.6% for Fable 5.2 and 75.3% for Mythos 5.2. Anthropic also says the safeguards are as robust as those on 5.1 and produce fewer unnecessary refusals.

Independent testing is more mixed. On Zapier's task benchmark, Fable 5.2 at maximum effort reached a 51.4% pass rate, against 53.9% for Gemini 4 Carbon at high effort. The New Stack noted that even Anthropic's own table does not show Fable 5.2 ahead on every test.

The price cut of 20% on both input and output follows Fable 5.1's $10 and $50 rates. Anthropic still advises customers to start with Opus 5.5 and move up only when a task requires it. That advice concedes that most workloads do not need the top tier.