AIME 2025 (avg@8)
Math · 2025-02-12
AIME 2025 accuracy averaged across eight sampled solutions per problem, separated from single-attempt results.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 97.1 |
| Muse Spark | 97.1 |
| GPT-5.2 | 96.7 |
| GPT-5.4 | 96.3 |
| Opus 4.7 | 95.8 |
| Gemini 3 Flash Preview | 95.4 |
| GPT-5.4 Mini | 95.4 |
| Gemini 3 Pro Preview | 94.2 |
| Grok 4.20 | 94.2 |
| Opus 4.6 | 93.8 |
| Kimi K2.5 | 93.8 |
| Qwen3.6 Plus (2026-04-02) | 93.8 |
| GPT-5.1 | 92.9 |
| GPT-5 | 92.6 |
| GPT-5.4 Nano | 92.1 |
| gpt-oss-120b | 92.1 |
| GLM-4.7 | 91.7 |
| GLM-5.1 | 91.7 |
| Qwen3.5-Flash | 91.7 |
| Opus 4.5 | 91.3 |
| Sonnet 4.6 | 91.3 |
| GLM 4.6 | 90.8 |
| Grok 4.1 Fast | 90.4 |
| MiniMax M2.7 | 90.4 |
| GLM-5 | 90.0 |
| GPT-5 Mini | 90.0 |
| Grok 4 Fast | 88.3 |
| MiniMax M2.5 | 88.3 |
| Grok 4 | 87.8 |
| GLM-4.5 | 86.3 |
| Sonnet 4.5 | 85.7 |
| gpt-oss-20b | 85.4 |
| Kimi K2 Thinking | 85.0 |
| Gemini 2.5 Pro Experimental 03-25 | 84.2 |
| o3-mini | 83.8 |
| Qwen3.5 Plus (2026-02-15) | 83.8 |
| o3 | 83.3 |
| Qwen3-235B-A22B | 83.3 |
| O4 Mini | 82.7 |
| Haiku 4.5 | 81.7 |