Expert-level Question Answering on GPQA Main
40.2AccuracyQwen3.5-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3.5-4BParameters=4B2026.05 | 40.2 | |
| Qwen3-4BParameters=4B2026.05 | 36.8 | |
| Mellum 2Parameters=2.5B/12B2026.05 | 35 | |
| Qwen2.5-7BParameters=7B2026.05 | 34.2 | |
| OLMo-3-7BParameters=7B2026.05 | 27.9 |