ARC-AGI-2 (Semi-Private)
ARC-AGI-2 (Semi-Private) is run by ARC Prize Foundation. It has ranked 232 models, of which the catalogue holds 64, scoring from 0 to 95 on Score (%).
measured by ARC Prize Foundation · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 2ndof 232 | GPT 6 Astra | Score (%) (max) | 95 |
| 4thof 232 | GPT-5.6 Sol | Score (%) (max) | 92.5 |
| 7thof 232 | Claude Opus 5 | Score (%) (max) | 90.42 |
| 9thof 232 | Claude Fable 5.1 | Score (%) (max) | 90 |
| 11thof 232 | Claude Fable 5 | Score (%) (max) | 89.17 |
| 21stof 232 | GPT-5.5 Pro | Score (%) (high) | 84.58 |
| 22ndof 232 | Gemini 3.7 Flash | Score (%) (high) | 84.58 |
| 24thof 232 | GPT-5.6 Terra | Score (%) (max) | 83.9 |
| 25thof 232 | GPT-5.4 Pro | Score (%) (xhigh) | 83.33 |
| 29thof 232 | Gemini 3.1 Pro Preview | Score (%) | 77.08 |
| 31stof 232 | Claude Opus 4.7 | Score (%) (max) | 75.83 |
| 33rdof 232 | GPT-5.4 | Score (%) (xhigh) | 73.95 |
| 35thof 232 | Claude Opus 4.8 | Score (%) (high) | 72.08 |
| 36thof 232 | Gemini 3.5 Flash | Score (%) (high) | 72.08 |
| 39thof 232 | Claude Opus 4.6 | Score (%) (high) | 69.17 |
| 46thof 232 | Grok 4.6 | Score (%) (xhigh) | 67.08 |
| 54thof 232 | DeepSeek V4 Flash (0731) | Score (%) (max) | 61.39 |
| 56thof 232 | DeepSeek V4 Pro 0813 | Score (%) (max) | 61.25 |
| 57thof 232 | Claude Sonnet 4.6 | Score (%) (high) | 60.42 |
| 58thof 232 | Kimi K3 | Score (%) (max) | 60.42 |
| 59thof 232 | Gemini 3.6 Flash | Score (%) (high) | 60.42 |
| 61stof 232 | GPT-5.6 Luna | Score (%) (max) | 59.58 |
| 69thof 232 | GPT-5.2 Pro | Score (%) (high) | 54.16 |
| 73rdof 232 | Grok 4.5 | Score (%) (high) | 52.64 |
| 78thof 232 | Gemini 3 Deep Think | Score (%) | 45.14 |
| 82ndof 232 | Inkling Small | Score (%) (xhigh) | 40.14 |
| 86thof 232 | Inkling | Score (%) | 36.53 |
| 92ndof 232 | gemini-3-pro-preview | Score (%) | 31.11 |
| 103rdof 232 | GLM 5.2 | Score (%) | 22.78 |
| 104thof 232 | GPT-5.4 mini | Score (%) (xhigh) | 18.9 |
| 106thof 232 | GPT-5 Pro | Score (%) | 18.33 |
| 107thof 232 | GPT 5.1 Thinking | Score (%) (high) | 17.64 |
| 115thof 232 | Kimi K2.5 | Score (%) | 11.81 |
| 116thof 232 | Gemini 3.5 Flash-Lite | Score (%) (high) | 10.28 |
| 117thof 232 | GPT-5 | Score (%) (high) | 9.86 |
| 121stof 232 | Claude Opus 4 | Score (%) | 8.61 |
| 128thof 232 | o3 | Score (%) (high) | 6.53 |
| 131stof 232 | o4-mini | Score (%) (high) | 6.11 |
| 132ndof 232 | Claude Sonnet 4 | Score (%) | 5.93 |
| 134thof 232 | GPT-5.4 nano | Score (%) (xhigh) | 5.69 |
| 138thof 232 | o3-pro | Score (%) (high) | 4.86 |
| 140thof 232 | MiniMax M2.5 | Score (%) | 4.86 |
| 141stof 232 | GLM-5 | Score (%) | 4.86 |
| 144thof 232 | GPT-5 mini | Score (%) (high) | 4.44 |
| 149thof 232 | DeepSeek v3.2 | Score (%) | 4.03 |
| 152ndof 232 | Gemini 2.5 Pro | Score (%) | 3.75 |
| 153rdof 232 | Claude Sonnet 4.5 | Score (%) | 3.75 |
| 158thof 232 | o3-mini | Score (%) (high) | 2.99 |
| 163rdof 232 | GPT-5 nano | Score (%) (high) | 2.61 |
| 181stof 232 | Gemini 2.5 Flash | Score (%) | 1.69 |
| 188thof 232 | Gemini 2.0 Flash | Score (%) | 1.3 |
| 190thof 232 | Codex mini | Score (%) | 1.27 |
| 193rdof 232 | Qwen3-235B-A22B-Instruct-2507 | Score (%) | 1.25 |
| 194thof 232 | Claude Haiku 4.5 | Score (%) | 1.25 |
| 197thof 232 | DeepSeek-R1 (0528) | Score (%) | 1.12 |
| 202ndof 232 | o1-mini | Score (%) | 83 |
| 211thof 232 | GPT-4.1 fine-tuned | Score (%) | 42 |
| 212thof 232 | Grok 3 Mini | Score (%) (low) | 42 |
| 216thof 232 | GPT-4o (2024-11-20) | Score (%) | 0 |
| 217thof 232 | GPT-4o-mini (2024-07-18) | Score (%) | 0 |
| 220thof 232 | GPT-4.1 nano | Score (%) | 0 |
| 221stof 232 | GPT-4.1 mini | Score (%) | 0 |
| 224thof 232 | Grok 3 | Score (%) | 0 |
| 225thof 232 | Magistral Small 2506 | Score (%) | 0 |
Read from the board on 2026-09-16