Scientific Reasoning on GPQA Diamond (p@1, p@16)
91.94P@16LED
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LEDModel=Qwen3-30B-A3B-Thinking2026.02 | 91.94 | 71.09 | |
| DoLaModel=Qwen3-30B-A3B-Thinking2026.02 | 87.88 | 71.53 | |
| ST-GModel=Qwen3-30B-A3B-Thinking2026.02 | 87.88 | 72.06 | |
| DoLaModel=Qwen3-4B-Thinking2026.02 | 87.37 | 64.36 | |
| DoLaModel=MiMo-7B-RL2026.02 | 86.87 | 51.86 | |
| CoTModel=MiMo-7B-RL2026.02 | 86.36 | 51.36 | |
| CoTModel=Qwen3-30B-A3B-Thinking2026.02 | 86.36 | 70.77 | |
| ST-GModel=MiMo-7B-RL2026.02 | 85.56 | 52.18 | |
| ST-GModel=Qwen3-4B-Thinking2026.02 | 85.35 | 64.61 | |
| LEDModel=Qwen3-4B-Thinking2026.02 | 85.35 | 64.9 | |
| LEDModel=MiMo-7B-RL2026.02 | 85.35 | 52.18 | |
| CoTModel=Qwen3-4B-Thinking2026.02 | 83.33 | 63.71 | |
| STModel=MiMo-7B-RL2026.02 | 80.3 | 51.99 | |
| STModel=Qwen3-30B-A3B-Thinking2026.02 | 78.28 | 71.88 | |
| STModel=Qwen3-4B-Thinking2026.02 | 77.78 | 64.9 | |
| Ministral 3Model Size=14B2026.01 | 71.2 | — | |
| Qwen3-VLModel Size=8B2026.01 | 67.1 | — | |
| Ministral 3Model Size=8B2026.01 | 66.8 | — | |
| Qwen 3Model Size=14B2026.01 | 66.3 | — | |
| Qwen3-VLModel Size=4B2026.01 | 60.1 | — | |
| Ministral 3Model Size=3B2026.01 | 53.4 | — |