Skip to content

AnthropicFrontier

Claude Fable 5.2

Same model as Mythos 5.2 with cyber and biology safeguards, at $8 in / $40 out per million tokens (less than Fable 5.1). Level with GPT-6.2 Astra at the top of the Epoch Capabilities Index (173.5, revised from 174.2) and first on ARC-AGI-3 (81.2%).

Announced
Access
Public

Also released as

  • Claude Mythos 5.2 · Limited preview · 9 Nov 2026

    Same model as Fable 5.2 with fewer cyber and biology safeguards. Only for vetted cyber-defence and life-science organisations.

Benchmark scores

TestScoreBest on the trackerHow it was measured
Epoch Capabilities Indexmany tests combined into one score174 pointsGPT-6.2 Astra 174 pointsIndependentlisted as "Claude Fable 5.2", CI 170.09 to 177.36 (Epoch revised it from 174.24 on 17 November)Run 9 Nov 2026
AA Intelligence Index v5about ten tests in one score64.6 pointsTop scoreIndependentv5.0.1; listed as Claude Fable 5.2 (Max, Default Fallback)Run 10 Nov 2026
Remote Labor Indexpaid freelance projects41%Top scoreIndependentFable 5.2Run 9 Nov 2026
FrontierSWE V2very long engineering jobs70%Top scoreIndependentproximus harness, mean@5 over 34 tasks, 20-hour budget, ±8.6 (as listed on frontierswe.com 10 Nov 2026; same figure appears in Anthropic's launch post)
APEX-Agentsprofessional office tasks80%Gemini 4 Carbon 84%IndependentMax effort, ±4.7 (read from the live Mercor page 10 Nov 2026; not yet in Epoch mirror)
ARC-AGI-3puzzle games, standard harnessTypical first-time human player: 100%81%Top scoreIndependentARC Prize verified, Standard harness, XHigh reasoning (79.3% at High, 65.7% at Max)Run 9 Nov 2026
FrontierMath T1–3hard maths problems95%Top scoreIndependentmax reasoning effort (Epoch AI run)Run 9 Nov 2026
GPQA DiamondPhD-level science questionsPhD experts: 70%Saturated94%Gemini 4 Argon 96%Independentmax reasoning effort (Epoch AI run)Run 9 Nov 2026
FrontierMath T4research-level maths problemsSaturated100%Top scoreIndependentmax reasoning effort (Epoch AI run)Run 9 Nov 2026
ARC-AGI-2visual puzzles, model aloneAverage person: 60%Saturated94%GPT-6 Astra 95%IndependentARC Prize verified, semi-private set, XHigh reasoningRun 9 Nov 2026
OTIS Mock AIMEcompetition mathsSaturated100%GPT-5.5 100%Independentmax reasoning effort (Epoch AI run)Run 9 Nov 2026

Compare every model on the same tests →

Release story

Also on these charts

Sources