MATH leaderboard
As of November 14, 2026, among the results listed here, Kimi-K2-Base has the highest reported MATH score, 70.2 (Base model evaluation results; EM), published by Moonshot AI, ahead of Meta Llama 4 Maverick at 61.2. 4 models have a MATH result in our records.
Scores as the model makers published them, not re-run by us; each row links to its announcement.
Best reported score
70.2
Reported by Moonshot AI, July 11, 2025 · Base model evaluation results; EM
4 models ranked · 2 makers
| Rank | Model | Score | Setting | Reported by | Date |
|---|---|---|---|---|---|
| 1 | Kimi-K2-Base | 70.2 | Base model evaluation results; EM | Moonshot AI | July 11, 2025 (model release) |
| 2 | Llama 4 Maverick | 61.2 | 4 | Meta Llama | April 5, 2025 (model release) |
| 3 | Llama 4 Scout | 50.3 | 4 | Meta Llama | April 5, 2025 (model release) |
| 4 | Llama 3.1 70B Instruct | 41.6 | 4 | Meta Llama | April 5, 2025 (model release) |
How much to trust these numbers
- Source
- Reported by the model makers themselves. Not re-run by LLMBenchmark or anyone independent.
- Settings
- Settings differ between makers. Check the Setting column before comparing two scores.
- Spread
- 70.2 to 41.6 across 4 models.
- Freshness
- Newest result published July 11, 2025; oldest April 5, 2025.
What MATH measures
MATH asks models to solve challenging competition mathematics problems, with scores indicating performance on those problems. Source: Measuring Mathematical Problem Solving With the MATH Dataset