Claude Mythos 5.2

Anthropic makes Claude, one of the AI systems that work on Spellwright's Intelligence dossiers.

Closed

Anthropic released Claude Mythos 5.2 on November 9, 2026, under Claude API ID claude-mythos-5-2. Anthropic says it is the same underlying model as Claude Fable 5.2 without certain cybersecurity and biology safeguards, and limits access to organizations verified through its verification programs.

Benchmark scores

Terminal-Bench-Science 0.1 · Reported by the model's maker, Nov 9, 2026Setting: Highest-available effort, per the table footnote; the effort level is not named for this row. Numeric reasoning and token budgets are not stated.75.3%#1 of 2 results
CVE Rediscovery · Reported by the model's maker, Nov 9, 2026Setting: Highest-available effort, per the table footnote; the effort level is not named for this row. Numeric reasoning and token budgets are not stated.89.6%#1 of 1 results
CyScenarioBench · Reported by the model's maker, Nov 9, 2026Setting: Highest-available effort, per the table footnote; the effort level is not named for this row. Numeric reasoning and token budgets are not stated.76.3%#1 of 1 results
De novo binder design · Reported by the model's maker, Nov 9, 2026Setting: Highest-available effort, per the table footnote; the effort level is not named for this row. Numeric reasoning and token budgets are not stated.85.1%#1 of 1 results

Access

Verified organizations only. Anthropic lists the model as active but usable only by organizations verified through its programs, such as the Cyber Verification Program and the Life Sciences Verification Program. Anthropic lists it as released and active, with verification required; it does not call it a preview. Platforms listed: Claude API, Amazon Bedrock, Google Cloud (Google managed and Anthropic managed) and Microsoft Foundry. Anthropic also says Claude Enterprise customers can reach it through Claude Security. No account entitlement or deployment was tested.