Benchmarks · math
AIME 2025
AIME 2025: competition math; accuracy.
47 models · 60 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
GPT-5.2 xhigh
All results
60 rows| # | Model | Effort | AIME 2025 |
|---|---|---|---|
| 1 | GPT-5.2 | xhigh | 99.0% |
| 2 | GPT-5 Codex | high | 98.7% |
| 3 | GPT-5.2 | medium | 96.7% |
| 4 | GPT-5.1 Codex | high | 95.7% |
| 5 | GLM 4.7 | 95.0% | |
| 6 | GPT-5 | high | 94.3% |
| 7 | GPT-5.1 | high | 94.0% |
| 8 | gpt-oss-120b | high | 93.4% |
| 9 | Grok 4 | 92.7% | |
| 10 | GPT-5.1 Codex Mini | high | 91.7% |
| 11 | GPT-5 | medium | 91.7% |
| 12 | GPT-5 Mini | high | 90.7% |
| 13 | o4 mini | high | 90.7% |
| 14 | gpt-oss-20b | high | 89.3% |
| 15 | o3 | 88.3% | |
| 16 | Gemini 2.5 Pro | 87.7% | |
| 17 | GPT-5 Mini | medium | 85.0% |
| 18 | GPT-5 Nano | high | 83.7% |
| 19 | GPT-5 | low | 83.0% |
| 20 | MiniMax M2.1 | 82.7% | |
| 21 | GLM 4.5 Air | 80.7% | |
| 22 | Qwen 3 Max | none | 80.7% |
| 23 | GPT-5 Nano | medium | 78.3% |
| 24 | MiniMax M2 | 78.3% | |
| 25 | GLM 4.5 | 73.7% | |
| 26 | MiMo V2 Flash | none | 67.7% |
| 27 | gpt-oss-120b | low | 66.7% |
| 28 | Opus 4.5 | none | 62.7% |
| 29 | gpt-oss-20b | low | 62.3% |
| 30 | Gemini 2.5 Flash | none | 60.3% |