ARC-AGI-1
General QA · 2019-11-05
Concept-learning benchmark for abstraction and program-like reasoning from input-output examples.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 98.0 |
| Claude Opus 5 | 97.5 |
| GPT-5.6 Sol | 97.5 |
| GPT-5.5 Pro | 96.5 |
| GPT-5.6 Terra | 96.5 |
| Gemini 3 Deep Think | 96.0 |
| GPT-5.5 | 95.0 |
| GPT-5.4 Pro | 94.5 |
| Kimi K3 | 94.5 |
| Opus 4.6 | 94.0 |
| Opus 4.7 | 93.5 |
| GPT-5.4 | 92.7 |
| Opus 4.8 | 92.5 |
| Gemini 3.5 Flash | 92.5 |
| GPT-5.2 Pro | 90.5 |
| Grok 4.20 | 89.5 |
| GPT-5.6 Luna | 88.0 |
| Grok 4.5 | 87.2 |
| Sonnet 4.6 | 86.5 |
| Gemini 3 Flash Preview | 84.7 |
| Inkling-Small | 84.0 |
| Opus 4.5 | 80.0 |
| Inkling | 79.5 |
| GLM-5.2 | 77.0 |
| Gemini 3 Pro Preview | 75.0 |
| GPT-5.2 | 72.7 |
| GPT-5 Pro | 70.2 |
| Grok 4 | 66.7 |
| Kimi K2.5 | 65.3 |
| Sonnet 4.5 | 63.7 |
| GPT-5.4 Mini | 63.7 |
| MiniMax M2.5 | 63.7 |
| DeepSeek-V3.2 | 57.0 |
| o3-pro | 57.0 |
| GPT-5 Mini | 54.3 |
| GPT-5.4 Nano | 51.5 |
| Grok 4 Fast | 48.5 |
| Haiku 4.5 | 47.7 |
| GLM-5 | 44.7 |
| GPT-5 | 44.0 |