Division
Reasoning Division
Ranked by GPQA Diamond: graduate-level science that resists lookup.
| # | Model | Maker | Division metric | League Score |
|---|---|---|---|---|
| 1 | GPT-6 Astra | OpenAI | 96.0% GPQA Diamond | 97 |
| 2 | Gemini 3.8 Flash | Google DeepMind | 95.0% GPQA Diamond | 63 |
| 3 | GPT-5.6 Sol | OpenAI | 94.6% GPQA Diamond | 82 |
| 4 | Gemini 3.1 Pro Preview | Google DeepMind | 94.3% GPQA Diamond | 52 |
| 5 | Claude Fable 5.1 | Anthropic | 93.7% GPQA Diamond | 97 |
| 6 | Kimi K3 | Moonshot AI | 93.5% GPQA Diamond | 63 |
| 7 | Claude Opus 5 | Anthropic | 93.2% GPQA Diamond | 92 |
| 8 | GPT-5.6 Terra | OpenAI | 92.9% GPQA Diamond | 74 |
| 9 | Qwen3.8-Max | Alibaba Qwen | 92.6% GPQA Diamond | 62 |
| 10 | GPT-5.6 Luna | OpenAI | 92.3% GPQA Diamond | 60 |
| 11 | DeepSeek V4.1 Flash | DeepSeek | 90.9% GPQA Diamond | 75 |
| 12 | Kimi K2.6 | Moonshot AI | 90.5% GPQA Diamond | 70 |
| 13 | Grok 4.3 | xAI (SpaceXAI) | 90.1% GPQA Diamond | 74 |
| 14 | DeepSeek V4 Pro (0813) | DeepSeek | 90.1% GPQA Diamond | 63 |
| 15 | Qwen3.5-397B-A17B | Alibaba Qwen | 88.4% GPQA Diamond | 61 |
| 16 | Nemotron 3 Ultra 550B-A55B | NVIDIA | 87.0% GPQA Diamond | 57 |
| 17 | Gemma 4 31B | Google DeepMind | 84.3% GPQA Diamond | 49 |
| 18 | Gemma 4 26B A4B | Google DeepMind | 82.3% GPQA Diamond | 40 |
| 19 | Nemotron 3 Nano 30B-A3B | NVIDIA | 73.0% GPQA Diamond | 0 |
Division tables regenerate from the model database on every update. Scoring rules: methodology.