ARC-AGI-2
ARC Prize Foundation · verified-8d1a893f26ff · Verified ARC Prize results on the harder ARC-AGI-2 abstract reasoning benchmark.
Model ranking
Publisher benchmark →Source | |||||||
|---|---|---|---|---|---|---|---|
| #1 | GPT-5.6 SolOpenAI | 92.50 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #2 | Claude Opus 5Anthropic | 90.40 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #3 | Claude Fable 5Anthropic | 89.20 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #4 | GPT-5.6 TerraOpenAI | 83.90 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #5 | DeepSeek-V4-FlashDeepSeek | 61.40 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #6 | DeepSeek-V4-ProDeepSeek | 61.30 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #7 | Kimi-K3Moonshot AI | 60.40 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
| #8 | GPT-5.6 LunaOpenAI | 59.50 | — | — | Max reasoning; ARC Prize verified leaderboardunknown | Original sourcewinner eligible | Third-party benchmark |
8 models ranked by the best winner-eligible current-version score when available, otherwise the best publisher-artifact score; highest first. Observed Aug 29, 2026.
Ranking protocol
verified score · percent. Each model appears once; verified winner-eligible runs take precedence, and missing models are not imputed.
Acquisition coverage
8 published models from 8 source rows; 0 source identities remain quarantined.
Primary evidence
Publisher artifacts
ARC-AGI-2 methodologyARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 GPT-5.6 Sol (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 Claude Opus 5 (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 Claude Fable 5 (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 GPT-5.6 Terra (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 DeepSeek V4 Flash 0731 (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 DeepSeek V4 Pro 0813 (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 Kimi K3 (Max) verified resultARC Prize Foundation · observed Aug 29, 2026ARC-AGI-2 GPT-5.6 Luna (Max) verified resultARC Prize Foundation · observed Aug 29, 2026