Arc agi 2 — Epoch AI
Arc agi 2 — Epoch AI is run by Epoch AI. It has ranked 227 models, of which the catalogue holds 74, scoring from 0 to 95.
measured by Epoch AI · the board itself
| Place | Model | Metric | Score |
|---|---|---|---|
| 1stof 227 | GPT 6 Astra | Score (max) | 95 |
| 3rdof 227 | GPT-5.6 Sol | Score (max) | 92.5 |
| 6thof 227 | Claude Opus 5 | Score (max) | 90.42 |
| 9thof 227 | Claude Fable 5.1 | Score (max) | 90 |
| 11thof 227 | Claude Fable 5 | Score (max) | 89.167 |
| 20thof 227 | GPT-5.5 | Score (xhigh) | 85 |
| 22ndof 227 | Gemini 3.7 Flash | Score (high) | 84.583 |
| 23rdof 227 | Gemini 3 Deep Think | Score | 84.58 |
| 24thof 227 | GPT-5.5 Pro | Score (high) | 84.58 |
| 27thof 227 | GPT-5.6 Terra | Score (max) | 83.9 |
| 28thof 227 | GPT-5.4 Pro | Score (xhigh) | 83.33 |
| 33rdof 227 | Gemini 3.1 Pro Preview | Score | 77.1 |
| 36thof 227 | Claude Opus 4.7 | Score (max) | 75.83 |
| 40thof 227 | GPT-5.4 | Score (xhigh) | 73.95 |
| 42ndof 227 | Gemini 3.5 Flash | Score (high) | 72.08 |
| 43rdof 227 | Claude Opus 4.8 | Score (high) | 72.08 |
| 46thof 227 | Claude Opus 4.6 | Score | 69.17 |
| 52ndof 227 | Grok 4.6 | Score (xhigh) | 67.083 |
| 56thof 227 | Grok 4.20 | Score | 65.14 |
| 62ndof 227 | DeepSeek V4 Flash (0731) | Score (max) | 61.389 |
| 64thof 227 | DeepSeek V4 Pro 0813 | Score (max) | 61.25 |
| 65thof 227 | Claude Sonnet 4.6 | Score (high) | 60.42 |
| 66thof 227 | Kimi K3 | Score (max) | 60.417 |
| 67thof 227 | Gemini 3.6 Flash | Score (high) | 60.417 |
| 70thof 227 | GPT-5.6 Luna | Score (max) | 59.54 |
| 76thof 227 | GPT-5.2 Pro | Score (high) | 54.16 |
| 78thof 227 | GPT-5.2 | Score (xhigh) | 52.91 |
| 81stof 227 | Grok 4.5 | Score (high) | 52.64 |
| 90thof 227 | Inkling Small | Score (xhigh) | 40.139 |
| 92ndof 227 | Claude Opus 4.5 | Score | 37.64 |
| 94thof 227 | Inkling | Score | 36.528 |
| 95thof 227 | Gemini 3 Flash Preview | Score | 33.61 |
| 99thof 227 | gemini-3-pro-preview | Score | 31.11 |
| 106thof 227 | GLM 5.2 | Score | 22.775 |
| 107thof 227 | GPT-5.4 mini | Score (xhigh) | 18.9 |
| 110thof 227 | GPT-5 Pro | Score | 18.33 |
| 111thof 227 | GPT-5.1 | Score (high) | 17.64 |
| 115thof 227 | Claude Sonnet 4.5 | Score | 13.61 |
| 119thof 227 | Kimi K2.5 | Score | 11.81 |
| 120thof 227 | Gemini 3.5 Flash-Lite | Score (high) | 10.278 |
| 121stof 227 | GPT-5 | Score (high) | 9.86 |
| 124thof 227 | Claude Opus 4 | Score | 8.61 |
| 130thof 227 | o3 | Score (high) | 6.53 |
| 133rdof 227 | o4-mini | Score (high) | 6.11 |
| 134thof 227 | Claude Sonnet 4 | Score | 5.93 |
| 136thof 227 | GPT-5.4 nano | Score (xhigh) | 5.69 |
| 138thof 227 | Grok 4 Fast | Score | 5.28 |
| 142ndof 227 | o3-pro | Score (high) | 4.86 |
| 143rdof 227 | Gemini 2.5 Pro | Score | 4.86 |
| 144thof 227 | MiniMax M2.5 | Score | 4.86 |
| 145thof 227 | GLM-5 | Score | 4.86 |
| 147thof 227 | GPT-5 mini | Score (high) | 4.44 |
| 151stof 227 | Claude Haiku 4.5 | Score | 4.03 |
| 152ndof 227 | DeepSeek v3.2 | Score | 4.03 |
| 157thof 227 | o3-mini | Score (high) | 2.99 |
| 162ndof 227 | GPT-5 nano | Score (high) | 2.61 |
| 163rdof 227 | Gemini 2.5 Flash | Score | 2.54 |
| 184thof 227 | gemini-2.0-flash-001 | Score | 1.3 |
| 185thof 227 | DeepSeek-R1 | Score | 1.3 |
| 187thof 227 | Codex mini | Score | 1.27 |
| 189thof 227 | Qwen3-235B-A22B-Instruct-2507 | Score | 1.25 |
| 193rdof 227 | DeepSeek-R1 (0528) | Score | 1.12 |
| 195thof 227 | Claude 3.7 Sonnet | Score | 90 |
| 200thof 227 | o1-mini | Score | 83 |
| 206thof 227 | GPT-4.1 fine-tuned | Score | 42 |
| 207thof 227 | Grok 3 Mini | Score (low) | 42 |
| 211thof 227 | GPT-4o-mini (2024-07-18) | Score | 0 |
| 212thof 227 | GPT-4.1 nano | Score | 0 |
| 214thof 227 | Llama-4-Maverick-17B-128E-Instruct | Score | 0 |
| 216thof 227 | GPT-4.1 mini | Score | 0 |
| 217thof 227 | Grok 3 | Score | 0 |
| 219thof 227 | Magistral Small 2506 | Score | 0 |
| 221stof 227 | Gemini 2.5 Pro Preview 06-05 | Score | 0 |
| 224thof 227 | GPT-4o (2024-11-20) | Score | 0 |
Read from the board on 2026-09-16