Benchmark
LiveBench
Math benchmarks measure how well models solve maths problems. Scores are percentages.
- Models measured
- 71
- Scores recorded
- 87
- Sources
- 2
- IQ weight
- 2.39
Score distribution
- 72.4%
- Mean
- 74.3%
- Median
- 9.1%
- Std dev
- 66.1%
- 10th percentile
- 79.5%
- 90th percentile
- 29.6%
- Min
This benchmark has a weight of 2.39 for chat models, measured over 70 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 51 | GLM 5.2overall· LiveBench data | 73.4% |
| 52 | QwQ 32B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 73.1% |
| 53 | Claude Opus 4.5overall, 64k, high· LiveBench data | 72.9% |
| 54 | Claude Haiku 5.5overall, xhigh· LiveBench data | 72.9% |
| 55 | Inkling FP4overall, xhigh· LiveBench data | 72.9% |
| 56 | Mistral Large 4overall, high· LiveBench data | 72.8% |
| 57 | GPT-5.3 Codex· LLM Stats data | 72.8% |
| 58 | DeepSeek V4 Prooverall· LiveBench data | 72.6% |
| 59 | Gemini 3 Flash· LLM Stats data | 72.4% |
| 60 | Kimi K2.6· LLM Stats data | 72.2% |
| 61 | GPT 5.1high· LLM Stats data | 72% |
| 62 | Kimi K2.7 Code· LLM Stats data | 71.9% |
| 63 | GLM 5.3 Flashoverall· LiveBench data | 71.1% |
| 64 | Kimi K2.6overall, thinking· LiveBench data | 70.9% |
| 65 | Qwen3.6 Plus· LLM Stats data | 70.8% |
| 66 | GPT 5.4 Nanooverall, xhigh· LiveBench data | 70.8% |
| 67 | GLM 5.1· LLM Stats data | 70.2% |
| 68 | GPT 5.4 Nano· LLM Stats data | 70.1% |
| 69 | MiniMax M3· LLM Stats data | 70% |
| 70 | Kimi K2.5· LLM Stats data | 69.1% |
| 71 | Qwen3.6 Plusoverall· LiveBench data | 69% |
| 72 | Kimi K2.7 Codeoverall· LiveBench data | 68.8% |
| 73 | Nemotron 3 Ultraoverall· LiveBench data | 68.7% |
| 74 | Grok Build 0.1overall· LiveBench data | 68.6% |
| 75 | MiniMax M3overall· LiveBench data | 67.5% |
| 76 | o1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 67% |
| 77 | GPT-5.4 minioverall, xhigh· LiveBench data | 66.6% |
| 78 | DeepSeek V4 Flashoverall· LiveBench data | 66.1% |
| 79 | Qwen3.6 27Boverall· LiveBench data | 64.2% |
| 80 | Gemini 3.5 Flash Liteoverall, high· LiveBench data | 63.8% |
| 81 | Grok 4.3overall· LiveBench data | 63.3% |
| 82 | O1 Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52.3% |
| 83 | Qwen2.5 72B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 52.3% |
| 84 | Phi 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 47.6% |
| 85 | Nemotron 3.5 Lightning 30B A3Boverall· LiveBench data | 45.8% |
| 86 | Qwen2.5 7B Instruct· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 35.9% |
| 87 | Qwen2.5-Omni 7B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 29.6% |
Showing 51–87 of 87 results.
Data credits: Data by LLM Stats