LiveBench Reasoning category score (0–100). Evaluates logic puzzles, spatial reasoning, and causal analysis.
Source: LiveBench| Rank | Model | |
|---|---|---|
| #1 | Anthropic Claude Opus 4.6 | 88.7 |
| #2 | ChatGPT GPT-5.4 | 88.1 |
| #3 | Gemini 3.1 Pro | 84.0 |
| #4 | Anthropic Claude Sonnet 4.6 | 77.4 |
| #5 | Moonshot AI Kimi K2.5 | 76.0 |
| #6 | Qwen Qwen3.6 Plus | 75.8 |
| #7 | MiniMax MiniMax M2.7 | 74.8 |
| #8 | Z.ai GLM-5.1 | 72.5 |
| #9 | Gemini 2.5 Pro | 70.8 |
| #10 | Xiaomi MiMo-V2-Pro | 69.7 |
| #11 | Z.ai GLM-5 | 69.1 |
| #12 | Gemini 3.1 Flash Lite | 59.7 |
| #13 | Gemma 4 31B | 59.4 |
| #14 | MiniMax MiniMax M2.5 | 59.3 |
| #15 | ChatGPT GPT-5 Mini | 58.6 |
| #16 | Gemini 3 Flash | 49.2 |
| #17 | Anthropic Claude Opus 4.5 | 47.8 |
| #18 | Gemini 2.5 Flash | 44.6 |
| #19 | DeepSeek DeepSeek V3.2 | 44.3 |
| #20 | Gemini 2.5 Flash Lite | 43.3 |
| #21 | Anthropic Claude Sonnet 4.5 | 42.3 |
| #22 | ChatGPT GPT OSS 120B | 39.2 |
| #23 | ChatGPT GPT-5 Nano | 35.5 |
| #24 | Anthropic Claude Haiku 4.5 | 33.9 |
| #25 | ChatGPT GPT-5.4 Mini | 21.9 |
| #26 | ChatGPT GPT-5.4 Nano | 17.4 |