CMIMC 2025
Math · 2025-05-29
This MathArena track evaluates models on the CMIMC 2025 problem set. Published scores average correctness across observed problem attempts; the item count records distinct problems, not a single-run binomial denominator.
Top models (higher is better)
| Model | Score |
|---|---|
| DeepSeek-V3.2-Speciale | 95.6 |
| GPT-5.2 | 95.6 |
| Step 3.5 Flash | 95.6 |
| GPT-5.1 | 94.4 |
| GLM-5 | 93.1 |
| Kimi K2.5 | 90.6 |
| Kimi K2 Thinking | 90.6 |
| Gemini 3 Flash Preview | 90.0 |
| Gemini 3 Pro Preview | 90.0 |
| GPT-5 | 88.8 |
| GLM 4.6 | 87.5 |
| gpt-oss-120b | 85.0 |
| Grok 4 | 85.0 |
| Grok 4 Fast | 85.0 |
| DeepSeek-V3.2 | 84.4 |
| GPT-5 Mini | 84.4 |
| Grok 4.1 Fast | 84.4 |
| O4 Mini | 83.8 |
| DeepSeek-V3.1 | 82.5 |
| o3 | 82.5 |
| DeepSeek-V3.2-Exp | 76.3 |
| GPT-5 Nano | 74.4 |
| gpt-oss-20b | 73.8 |
| GLM-4.5 | 72.5 |
| Falcon H1R 7B | 71.9 |
| GLM 4.5 Air | 71.9 |
| DeepSeek-R1-0528 | 71.3 |
| Grok 3 Mini | 68.1 |
| K2-Think | 68.1 |
| Sonnet 4.5 | 66.3 |
| QED-Nano | 61.9 |
| Gemini 2.5 Pro | 58.1 |
| Gemini 2.5 Flash | 54.4 |