Benchmarks · math
MATH-500
MATH-500: competition math subset; accuracy.
54 models · 60 results · data as of 2026-11-06 · artificialanalysis.ai ↗
Leaderboard
GPT-5 high
All results
60 rows| # | Model | Effort | MATH-500 |
|---|---|---|---|
| 1 | GPT-5 | high | 99.4% |
| 2 | Grok 3 Mini | high | 99.2% |
| 3 | o3 | 99.2% | |
| 4 | GPT-5 | medium | 99.1% |
| 5 | Grok 4 | 99.0% | |
| 6 | o4 mini | high | 98.9% |
| 7 | GPT-5 | low | 98.7% |
| 8 | o3 mini | high | 98.5% |
| 9 | Qwen 3 235B-A22B | 98.4% | |
| 10 | Opus 4 | 98.2% | |
| 11 | Gemini 2.5 Flash | 98.1% | |
| 12 | MiniMax M1 | 98.0% | |
| 13 | GLM 4.5 | 97.9% | |
| 14 | o3 mini | 97.3% | |
| 15 | Kimi K2 | none | 97.1% |
| 16 | o1 | 97.0% | |
| 17 | Gemini 2.5 Pro | 96.7% | |
| 18 | GLM 4.5 Air | 96.5% | |
| 19 | Qwen 3 14B | 96.1% | |
| 20 | Qwen 3 32B | 96.1% | |
| 21 | Qwen 3 30B-A3B | 95.9% | |
| 22 | o1 mini | 94.4% | |
| 23 | DeepSeek V3 | none | 94.2% |
| 24 | Gemini 2.5 Flash | none | 93.2% |
| 25 | Gemini 2.5 Flash Lite | none | 92.6% |
| 26 | GPT-4.1 Mini | none | 92.5% |
| 27 | GPT-4.1 | none | 91.3% |
| 28 | Mistral Medium 3 | none | 90.7% |
| 29 | Llama 4 Maverick | none | 88.9% |
| 30 | Mistral Small 3.2 | none | 88.3% |