Skip to content

Benchmark

LiveBench Math

Math benchmarks measure how well models solve maths problems. Scores are percentages. Higher is better.

Models measured
42
Scores recorded
46
Sources
1
IQ weight
1.86

Score distribution

54.4%
Mean
57.8%
Median
19.4%
Std dev
34.5%
10th percentile
79%
90th percentile
15.7%
Min
15.7%94.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.86 for chat models, measured over 41 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT 5.1high· Epoch AI Benchmarking Hub data94.5%
2Gemini 2.5 Pro· Epoch AI Benchmarking Hub data90.2%
3DeepSeek R1· Epoch AI Benchmarking Hub data80.7%
4o1high· Epoch AI Benchmarking Hub data80.3%
5Claude Sonnet 3.7· Epoch AI Benchmarking Hub data79%
6QwQ 32B· Epoch AI Benchmarking Hub data77.8%
7o3 Minihigh· Epoch AI Benchmarking Hub data77.3%
8Gemini 2.0 Flash Experimentalthinking· Epoch AI Benchmarking Hub data75.9%
9DeepSeek V3· Epoch AI Benchmarking Hub data73.5%
10o3 Minimedium· Epoch AI Benchmarking Hub data72.4%
11gemini-exp· Epoch AI Benchmarking Hub data72.4%
12gemini-2.0-pro-exp-02-05· Epoch AI Benchmarking Hub data71%
13GPT-4.5 Preview· Epoch AI Benchmarking Hub data69.3%
14Gemini 2.0 Flash· Epoch AI Benchmarking Hub data65.6%
15o3 Minilow· Epoch AI Benchmarking Hub data63.1%
16O1 Minimedium· Epoch AI Benchmarking Hub data62%
17Gemini 2.0 Flash Experimental· Epoch AI Benchmarking Hub data60.4%
18DeepSeek R1 Distill Qwen 32B· Epoch AI Benchmarking Hub data59.4%
19Qwen2.5-Max· Epoch AI Benchmarking Hub data58.4%
20QwQ 32B Preview· Epoch AI Benchmarking Hub data58.3%
21DeepSeek R1 Distill Llama 70b· Epoch AI Benchmarking Hub data58.1%
22Gemini 2.0 Flash Lite· Epoch AI Benchmarking Hub data58.1%
23Learnlm 1.5 Pro Experimental· Epoch AI Benchmarking Hub data57.8%
24Gemini 2.0 Flash-Lite Preview· Epoch AI Benchmarking Hub data55.5%
25Gemma 3 27B IT· Epoch AI Benchmarking Hub data55.4%
26Grok 2· Epoch AI Benchmarking Hub data54.9%
27Claude 3.5 Sonnet· Epoch AI Benchmarking Hub data52.3%
28GPT 4o· Epoch AI Benchmarking Hub data49.5%
29Qwen 2.5 Coder 32B Instruct· Epoch AI Benchmarking Hub data46.6%
30Claude 3 Opus· Epoch AI Benchmarking Hub data43.6%
31Mistral Large2411· Epoch AI Benchmarking Hub data42.6%
32Llama 3.3 70B Instruct· Epoch AI Benchmarking Hub data42.2%
33Phi 4· Epoch AI Benchmarking Hub data42%
34Perplexity Sonar· Epoch AI Benchmarking Hub data41.6%
35Mistral Small· Epoch AI Benchmarking Hub data39.9%
36Mistral Small2503· Epoch AI Benchmarking Hub data39.4%
37Nova Prov1· Epoch AI Benchmarking Hub data38%
38Nova Litev1· Epoch AI Benchmarking Hub data36.7%
39GPT-4o mini· Epoch AI Benchmarking Hub data36.3%
40Claude Haiku 3.5· Epoch AI Benchmarking Hub data35.5%
41Nova Microv1· Epoch AI Benchmarking Hub data34.5%
42Gemma 2 27B IT· Epoch AI Benchmarking Hub data26.5%
43Gemma 2 9B IT· Epoch AI Benchmarking Hub data19.8%
44C4AI Command R 08 2024· Epoch AI Benchmarking Hub data19.4%
45Phi-3 Small 8K Instruct· Epoch AI Benchmarking Hub data17.6%
46Phi-3 Mini 4K Instruct· Epoch AI Benchmarking Hub data15.7%

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)