Math
LiveBench mathematics
Competition mathematics, symbolic work and problem solving.
Published by LiveBench under Apache-2.0. Compare it with other tests
Top 15
Each model at its best configuration, coloured by maker.
The best model from each maker
Each maker’s highest-scoring model on this test, and its rank.
| Maker | Best model | Score | Rank |
|---|---|---|---|
 Anthropic | Claude Opus 5.5 (max effort) | 97.1% | #1 of 63 |
 OpenAI | GPT-6.1 Sol (max) | 96.8% | #3 of 63 |
 xAI | Grok 4.7 (xhigh) | 95.7% | #12 of 63 |
 DeepSeek | DeepSeek V4 Pro 0813 | 95.1% | #13 of 63 |
 Google | Gemini 3.7 Flash (high) | 93.5% | #17 of 63 |
 Alibaba | Qwen3.8 Max | 91.3% | #24 of 63 |
 Z.ai | GLM-5.2 | 89.8% | #32 of 63 |
 NVIDIA | Nemotron 3 Ultra 550B A55B | 88.7% | #37 of 63 |
 Moonshot AI | Kimi K3 | 84.4% | #50 of 63 |
 MiniMax | MiniMax-M3 | 77.0% | #62 of 63 |
New top scores
Each time a different model took the top spot, since the site began keeping this record.
- Claude Opus 5.5 took the top spot at 97.1% from Claude Fable 5.1, which had led at 97.0%.
Every result
63 models, each once at its best configuration, highest score first. Models with the same score share a rank.
| Rank | Model | Maker | Score |
|---|---|---|---|
| #1 | Claude Opus 5.5 (max effort) | Anthropic | 97.1% |
| #2 | Claude Fable 5.1 (max effort) | Anthropic | 97.0% |
| #3 | GPT-6.1 Sol (max) | OpenAI | 96.8% |
| #4 | GPT-6 Astra (max) | OpenAI | 96.8% |
| #5 | Claude Sonnet 5.5 (xhigh effort) | Anthropic | 96.7% |
| #6 | GPT-6 Sol (max) | OpenAI | 96.4% |
| #7 | GPT-5.6 Sol (max) | OpenAI | 96.2% |
| #8 | Claude Fable 5 (max effort) | Anthropic | 96.0% |
| #9 | Muse Spark 1.3 (xhigh) |  Meta AI | 96.0% |
| #10 | GPT-5.5 (xhigh) | OpenAI | 95.9% |
| #11 | Claude Opus 5 (max effort) | Anthropic | 95.7% |
| #12 | Grok 4.7 (xhigh) | xAI | 95.7% |
| #13 | DeepSeek V4 Pro 0813 | DeepSeek | 95.1% |
| #14 | GPT-5.6 Terra (max) | OpenAI | 94.9% |
| #15 | Claude Opus 4.8 (max effort) | Anthropic | 94.3% |
| #16 | GPT-5.4 (xhigh) | OpenAI | 94.2% |
| #17 | Gemini 3.7 Flash (high) | Google | 93.5% |
| #18 | DeepSeek V4.1 Flash (max) | DeepSeek | 93.3% |
| #19 | GPT-5.2 (2025 12 11 high) | OpenAI | 93.2% |
| #20 | Claude Sonnet 5 (xhigh effort) | Anthropic | 92.9% |
| #21 | Claude Opus 4.7 (xhigh effort) | Anthropic | 92.8% |
| #22 | Grok 4.6 | xAI | 92.6% |
| #23 | Gemini 3.8 Flash (high) | Google | 91.6% |
| #24 | Qwen3.8 Max | Alibaba | 91.3% |
| #25 | Muse Spark 1.2 (xhigh) | Meta AI | 91.2% |
How it is scored
Arithmetic mean of the 4 published task scores in this category. Only complete results are ranked. Exact model and effort configurations are preserved.
Results from LiveBench (Apache-2.0), as published. Model names link to their pages where the site can match them; makers come from the model’s page, or else from the publisher.