Mathematical Problem Solving on AIME 2025 (Pass@1)
95Pass@1Gemini-3.0 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3.0 Protemplate=reason step by step2025.12 | 95 | |
| GPT-5 Hightemplate=reason step by step2025.12 | 94.6 | |
| Kimi-K2thinking mode=true, template=reason step by step2025.12 | 94.5 | |
| DeepSeek-V3.2thinking mode=true, template=reason step by step2025.12 | 93.1 | |
| Claude-4.5-Sonnettemplate=reason step by step2025.12 | 87 | |
| MiniMax M2template=reason step by step2025.12 | 78.3 | |
| Qwen3-8Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 67.3 | |
| QUESTA-Nemotron-1.5Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 62.29 | |
| DeepSeek-R1-Distill-32Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 51.8 | |
| Nemotron-1.5Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 49.5 | |
| Qwen3-1.7Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 36.8 | |
| DeepSeek-R1-Distill-1.5Bk (responses per question)=32, sampling temperature=0.7, top-p=0.952025.07 | 22.3 |