Claude Fable 5.2 sets new ARC-AGI-3 high score: 81.2% standard harness at $21.9K per run
scaling01 · x · 2026-11-10
ARC Prize published verified scores for Anthropic's Claude Fable 5.2 on ARC-AGI:
- ARC-AGI-3: 97.7% with provider adapter harness ($18.6K); 81.2% on the standard harness at $21.9K — a new high score for the standard setting
- ARC-AGI-2: 93.8% at $0.98/task
- ARC-AGI-1: 98.5% at $0.36/task
The leaderboard chart shows Fable 5.2's adapter configuration clearly ahead of GPT-6 Astra and GPT-6.1 Sol-Prov entries, though each run costs tens of thousands of dollars in inference.
Related event: Claude Fable 5.2 Sets New Record on ARC-AGI-3(2 posts)→
More from AGI Musings
- Candidate uses phone AI to answer interview questions live in viral video — TansuYegen · 2026-11-10
- Industrial Revolution's uneven gains offer a cautionary map for AI prosperity — NinaDSchick · 2026-11-10
- Marty Cagan: 'execute-first' PMs find it far harder to get hired in the AI era — rseroter · 2026-11-10
- Why Go to Conferences When AI Reads and Writes the Papers? — plopesresearch · 2026-11-10
- Punishing reasoning models for 'bad thoughts' teaches them to hide cheating: OpenAI's own finding — r0ck3t23 · 2026-11-10
- Steven Strogatz's new book 'Big Math': machine-powered math can tackle millions of unknowns — stevenstrogatz · 2026-11-10