BENCHMARK EXPLORER · math
HMMT February 2026
Challenging competition mathematics from the Harvard–MIT tournament.
What it measures
Reported accuracy on the February 2026 competition.
Interpretation caveats
Sampling budgets and papers must be kept distinct.
Original benchmarkPublished results
| # | Modalities | Compare | ||||
|---|---|---|---|---|---|---|
| 1 | Qwen | InputTextImageVideoOutputText | 262,144tokens | |||
| 2 | moonshotai | InputTextImageOutputText | 262,144tokens | |||
| 3 | zai-org | InputTextOutputText | 204,800tokens | |||
| 3 | stepfun-ai | InputTextOutputText | 262,144tokens | |||
| 4 | deepseek-ai | InputTextOutputText | 163,840tokens | |||
| — | openbmb | — | InputTextOutputNot documented | —Not documented | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | Anthropic | — | InputTextImageFileOutputText | 1,000,000tokens | ||
| — | nvidia | — | InputTextOutputImage | —Not documented | ||
| — | nvidia | — | InputTextOutputImage | —Not documented | ||
| — | deepseek-ai | — | InputTextOutputText | 64,000tokens | ||
| — | deepseek-ai | — | InputTextOutputText | 163,840tokens | ||
| — | deepseek-ai | — | InputTextOutputText | 163,840tokens | ||
| — | deepseek-ai | — | InputTextOutputText | 163,840tokens | ||
| — | HelpingAI | — | InputTextOutputNot documented | —Not documented | ||
| — | tiiuae | — | InputTextOutputNot documented | —Not documented | ||
| — | zai-org | — | InputTextOutputText | 204,800tokens | ||
| — | zai-org | — | InputTextOutputText | 200,000tokens | ||
| — | — | InputTextOutputText | —Not documented | |||
| — | — | InputAudioFileImageTextVideoOutputText | 1,048,576tokens | |||
| — | — | InputTextImageVideoFileAudioOutputText | 1,048,576tokens | |||
| — | — | InputTextImageVideoFileAudioOutputText | 1,048,576tokens | |||
| — | — | InputAudioFileImageTextVideoOutputText | 1,048,576tokens | |||
| — | thinky | — | InputTextImageAudioOutputText | 524,288tokens | ||
| — | jdopensource | — | InputTextOutputNot documented | —Not documented | ||
| — | LGAI-EXAONE | — | InputTextOutputNot documented | —Not documented | ||
| — | Not reported | — | InputTextOutputText | —Not documented | ||
| — | Moonshot | — | InputTextImageVideoOutputText | 1,048,576tokens | ||
| — | moonshotai | — | InputTextOutputText | 262,144tokens | ||
| — | moonshotai | — | InputTextOutputText | 262,144tokens | ||
| — | LiquidAI | — | InputTextOutputNot documented | —Not documented | ||
| — | meta-llama | — | InputTextOutputText | 131,072tokens | ||
| — | meituan-longcat | — | InputTextOutputNot documented | —Not documented | ||
| — | meituan-longcat | — | InputTextOutputNot documented | —Not documented | ||
| — | XiaomiMiMo | — | InputTextOutputNot documented | —Not documented | ||
| — | MiniMaxAI | — | InputTextOutputText | 204,800tokens | ||
| — | MiniMaxAI | — | InputTextOutputText | 204,800tokens | ||
| — | MiniMaxAI | — | InputTextOutputText | 204,800tokens | ||
| — | miromind-ai | — | InputTextOutputNot documented | —Not documented | ||
| — | miromind-ai | — | InputTextOutputNot documented | —Not documented | ||
| — | Meta | — | InputTextOutputText | —Not documented | ||
| — | Meta | — | InputTextImageVideoFileOutputText | 1,048,576tokens | ||
| — | Meta | — | InputTextImageVideoFileOutputText | 1,048,576tokens |