LiveBench Math category score (0–100). Evaluates competition-level math problems, proofs, and numerical computation.
Source: LiveBench| Rank | Model | |
|---|---|---|
| #1 | ChatGPT GPT-5.4 | 94.2 |
| #2 | Gemini 3.1 Pro | 91.0 |
| #3 | Anthropic Claude Opus 4.6 | 89.3 |
| #4 | Z.ai GLM-5.1 | 84.9 |
| #5 | Moonshot AI Kimi K2.5 | 84.9 |
| #6 | Qwen Qwen3.6 Plus | 83.7 |
| #7 | Z.ai GLM-5 | 83.5 |
| #8 | MiniMax MiniMax M2.7 | 80.5 |
| #9 | Anthropic Claude Sonnet 4.6 | 80.4 |
| #10 | MiniMax MiniMax M2.5 | 77.4 |
| #11 | Xiaomi MiMo-V2-Pro | 77.0 |
| #12 | ChatGPT GPT-5 Mini | 74.4 |
| #13 | Gemma 4 31B | 73.9 |
| #14 | Gemini 3.1 Flash Lite | 73.6 |
| #15 | ChatGPT GPT OSS 120B | 68.9 |
| #16 | Gemini 2.5 Flash | 68.8 |
| #17 | Gemini 2.5 Pro | 68.3 |
| #18 | Gemini 3 Flash | 68.1 |
| #19 | ChatGPT GPT-5 Nano | 64.7 |
| #20 | Anthropic Claude Opus 4.5 | 64.1 |
| #21 | DeepSeek DeepSeek V3.2 | 64.0 |
| #22 | Anthropic Claude Sonnet 4.5 | 62.6 |
| #23 | Gemini 2.5 Flash Lite | 61.0 |
| #24 | Anthropic Claude Haiku 4.5 | 58.0 |
| #25 | ChatGPT GPT-5.4 Mini | 37.0 |
| #26 | ChatGPT GPT-5.4 Nano | 36.0 |