ARC-AGI-2 (Public Eval)
ARC-AGI-2 (Public Eval) is run by ARC Prize Foundation. It has ranked 216 models, of which the catalogue holds 60, scoring from 0 to 99.17.
measured by ARC Prize Foundation · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 2ndof 216 | Claude Fable 5.1 | Score (%) (xhigh) | 99.17 |
| 4thof 216 | GPT 6 Astra | Score (%) (xhigh) | 97.92 |
| 7thof 216 | Claude Opus 5 | Score (%) (max) | 97.06 |
| 9thof 216 | Claude Fable 5 | Score (%) (max) | 96.39 |
| 13thof 216 | GPT-5.6 Sol | Score (%) (max) | 93.75 |
| 16thof 216 | GPT-5.4 Pro | Score (%) (xhigh) | 92.21 |
| 18thof 216 | GPT-5.6 Terra | Score (%) (max) | 91.32 |
| 20thof 216 | GPT-5.5 Pro | Score (%) (xhigh) | 90.42 |
| 22ndof 216 | Gemini 3.1 Pro Preview | Score (%) | 88.07 |
| 27thof 216 | Gemini 3.7 Flash | Score (%) (high) | 85.28 |
| 28thof 216 | GPT-5.4 | Score (%) (xhigh) | 84.17 |
| 30thof 216 | Claude Opus 4.7 | Score (%) (max) | 82.23 |
| 35thof 216 | Claude Opus 4.6 | Score (%) (high) | 79.03 |
| 40thof 216 | Gemini 3.5 Flash | Score (%) (high) | 72.08 |
| 45thof 216 | Grok 4.6 | Score (%) (xhigh) | 68.33 |
| 48thof 216 | Claude Sonnet 4.6 | Score (%) (high) | 65.69 |
| 49thof 216 | Gemini 3.6 Flash | Score (%) (high) | 65.28 |
| 51stof 216 | GPT-5.6 Luna | Score (%) (max) | 64.17 |
| 52ndof 216 | DeepSeek V4 Flash (0731) | Score (%) (max) | 63.89 |
| 54thof 216 | DeepSeek V4 Pro 0813 | Score (%) (high) | 63.61 |
| 56thof 216 | Kimi K3 | Score (%) (max) | 61.67 |
| 64thof 216 | Grok 4.5 | Score (%) (high) | 58.19 |
| 70thof 216 | GPT-5.2 Pro | Score (%) (high) | 51.68 |
| 74thof 216 | Inkling Small | Score (%) (xhigh) | 39.31 |
| 78thof 216 | Inkling | Score (%) | 37.78 |
| 92ndof 216 | GLM 5.2 | Score (%) | 20.83 |
| 93rdof 216 | GPT 5.1 Thinking | Score (%) (high) | 18.33 |
| 94thof 216 | GPT-5.4 mini | Score (%) (xhigh) | 17.78 |
| 100thof 216 | GPT-5 Pro | Score (%) | 13.33 |
| 101stof 216 | Kimi K2.5 | Score (%) | 12.08 |
| 103rdof 216 | GPT-5 | Score (%) (high) | 9.58 |
| 107thof 216 | Gemini 3.5 Flash-Lite | Score (%) (high) | 7.92 |
| 110thof 216 | o4-mini | Score (%) (high) | 7.52 |
| 116thof 216 | GPT-5 mini | Score (%) (high) | 5.83 |
| 117thof 216 | MiniMax M2.5 | Score (%) | 5.42 |
| 118thof 216 | GLM-5 | Score (%) | 5.42 |
| 122ndof 216 | GPT-5.4 nano | Score (%) (high) | 5.14 |
| 128thof 216 | o3 | Score (%) (medium) | 4.49 |
| 129thof 216 | Claude Sonnet 4 | Score (%) | 4.44 |
| 133rdof 216 | o3-pro | Score (%) (high) | 3.89 |
| 134thof 216 | DeepSeek v3.2 | Score (%) | 3.89 |
| 135thof 216 | Claude Sonnet 4.5 | Score (%) | 3.75 |
| 140thof 216 | Claude Opus 4 | Score (%) | 3.33 |
| 157thof 216 | Gemini 2.5 Flash | Score (%) | 1.94 |
| 167thof 216 | Gemini 2.5 Pro | Score (%) | 1.11 |
| 168thof 216 | Qwen3-235B-A22B-Instruct-2507 | Score (%) | 83 |
| 176thof 216 | GPT-5 nano | Score (%) (high) | 30 |
| 177thof 216 | DeepSeek-R1 (0528) | Score (%) | 29 |
| 184thof 216 | GPT-4o (2024-11-20) | Score (%) | 0 |
| 185thof 216 | GPT-4o-mini (2024-07-18) | Score (%) | 0 |
| 188thof 216 | GPT-4.1 nano | Score (%) | 0 |
| 189thof 216 | GPT-4.1 mini | Score (%) | 0 |
| 190thof 216 | GPT-4.1 fine-tuned | Score (%) | 0 |
| 191stof 216 | o1-mini | Score (%) | 0 |
| 192ndof 216 | o3-mini | Score (%) (low) | 0 |
| 196thof 216 | Codex mini | Score (%) | 0 |
| 197thof 216 | Grok 3 | Score (%) | 0 |
| 198thof 216 | Grok 3 Mini | Score (%) (low) | 0 |
| 199thof 216 | Magistral Small 2506 | Score (%) | 0 |
| 207thof 216 | Claude Haiku 4.5 | Score (%) | 0 |
Read from the board on 2026-09-16