STEM Reasoning on MMLU-Redux 2.0
97.77Pass@1 AccuracyQwen3-30B-A3B (Thinking)
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-30B-A3B (Thinking)Sampling Strategy=4-sample mean2026.04 | 97.77 | |
| Gemini 2.5 FlashSampling Strategy=4-sample mean2026.04 | 96.85 | |
| GPT-5 MiniSampling Strategy=4-sample mean2026.04 | 96.4 | |
| GPT-OSS-120BSampling Strategy=4-sample mean2026.04 | 95.94 | |
| Nemotron 3 Nano 30B A3BSampling Strategy=4-sample mean2026.04 | 94.1 | |
| GPT-OSS-20BSampling Strategy=4-sample mean2026.04 | 93.32 | |
| Aryabhata 2Sampling Strategy=4-sample mean2026.04 | 92.92 | |
| GPT-5 NanoSampling Strategy=4-sample mean2026.04 | 88.47 |