AGI-Eval Community - General Reasoning: leaderboard
Metric: Accuracy (%). Source: agi-eval.cn. 141 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 97.82 |
| 2 | Gemini 3.1 Pro (Preview) | 97.68 |
| 3 | Claude Opus 4.6 | 97.66 |
| 4 | Qwen 3.7 Max | 97.56 |
| 5 | Gemini 3 Pro (Preview) | 97.51 |
| 6 | Kimi K3 | 97.5 |
| 7 | Ring-1T | 97.49 |
| 8 | Seed 2.1 Pro | 97.48 |
| 9 | Claude Opus 4 (Thinking) | 97.47 |
| 10 | Kimi K2.6 | 97.45 |
| 11 | Gemini 3.5 Flash | 97.45 |
| 12 | ERNIE 5.0 Thinking Preview | 97.41 |
| 13 | DeepSeek V4 Pro (Max) | 97.38 |
| 14 | Qwen 3 Next 80B A3B Instruct | 97.36 |
| 15 | GPT-5.5 | 97.34 |
Interactive version: theaggregate.ai/benchmark?slug=agi-eval-community-general-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-19.