LLM Stats (HMMT Feb 26): leaderboard
Problems from the February 2026 Harvard-MIT Mathematics Tournament, testing high-school olympiad mathematics across algebra, geometry, combinatorics, and number theory.
Metric: Score (%). Source: llm-stats.com. Source: qwen.ai. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.7 Max | 97.1 | #75 |
| 2 | Claude Opus 4.6 (Thinking) | 96.2 | #63 (Claude Opus 4.6) |
| 3 | DeepSeek V4 Pro (Max) | 95.2 | #102 (DeepSeek V4 Pro) |
| 4 | DeepSeek V4 Pro (Reasoning, Max Effort) | 95.2 | #102 (DeepSeek V4 Pro) |
| 5 | DeepSeek V4 Flash (Max) | 94.8 | #137 (DeepSeek V4 Flash) |
| 6 | Qwen 3.7 Plus | 92.9 | #92 |
| 7 | Kimi K2.6 (Thinking) | 92.7 | #104 (Kimi K2.6) |
| 8 | GLM-5.2 | 92.5 | #73 |
| 9 | GLM-5.1 (Thinking) | 89.4 | #97 (GLM-5.1) |
| 10 | Qwen 3.6 Plus | 87.8 | #120 |
| 11 | Ling-3.0-flash | 87 | #206 |
| 12 | Qwen 3.6 27B | 84.3 | #170 |
| 13 | Qwen 3.6 35B A3B | 83.6 | #226 |
| 14 | GLM-5.1 | 82.6 | #97 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-hmmt-feb-26 · How It Works · Data refreshed daily, snapshot 2026-10-20.