GraphARC: leaderboard
Metric: Average accuracy (self-reported). Source: benchmarklist.com. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 91 |
| 2 | O4 Mini | 87 |
| 3 | O3 Mini | 81 |
| 4 | O1 Mini | 58 |
| 5 | Qwen 3 14B | 37 |
| 6 | Qwen 3 32B | 34 |
| 7 | GPT-4.1 Nano | 12 |
| 8 | Llama 3.1 8B | 9 |
| 9 | Mistral-7B-v0.2 | 2 |
| 10 | OLMo 2 7B | 1 |
Interactive version: theaggregate.ai/benchmark?slug=grapharc · How It Works · Data refreshed daily, snapshot 2026-09-05.