HMMT Feb 2026
Math · 2026-02-19
This MathArena track evaluates models on the HMMT Feb 2026 problem set. Scores report the percentage of problems answered correctly.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Luna | 98.5 |
| GPT-5.5 | 98.5 |
| GPT-5.4 | 97.7 |
| GPT-5.2 | 97.0 |
| Kimi K3 | 97.0 |
| Opus 4.6 | 96.2 |
| Opus 4.8 | 95.5 |
| Gemini 3.5 Flash | 95.5 |
| Gemini 3.1 Pro Preview | 94.7 |
| Kimi K2.6 | 94.7 |
| Opus 4.7 | 93.9 |
| DeepSeek-V4-Flash | 93.9 |
| DeepSeek-V4-Pro | 93.9 |
| GLM-5.2 | 92.4 |
| Inkling-Small | 90.2 |
| Gemini 3.6 Flash | 89.4 |
| Gemini 3 Flash Preview | 89.4 |
| GLM-5.1 | 89.4 |
| Qwen3.5 397B A17B | 87.9 |
| Step 3.7 Flash | 87.9 |
| Kimi K2.5 | 87.1 |
| Gemini 3 Pro Preview | 86.4 |
| GLM-5 | 86.4 |
| Grok 4.1 Fast | 86.4 |
| Step 3.5 Flash | 86.4 |
| Inkling | 86.3 |
| Nemotron 3 Super 120B A12B | 84.8 |
| DeepSeek-V3.2 | 84.1 |
| MiMo-V2.5 | 82.6 |
| Qwen3.5 35B A3B | 81.8 |
| Qwen3.5-27B | 81.1 |
| Nemotron 3 Ultra 550B A55B | 78.8 |
| Qwen3 30B A3B Thinking 2507 | 78.8 |
| Qwen3.5-4B | 72.9 |
| Qwen3.5-9B | 71.2 |
| MiniMax M2.7 | 71.2 |
| Haiku 4.5 | 66.7 |
| QED-Nano | 64.4 |
| Gemini 3.5 Flash-Lite | 63.6 |
| Qwen3 4B Thinking 2507 | 53.0 |