Skip to content

Benchmark

LiveBench Reasoning

Scores are percentages.

Models measured
42
Scores recorded
46
Sources
1
IQ weight
2.06

Score distribution

54.1%
Mean
52.3%
Median
21.3%
Std dev
28.1%
10th percentile
87.8%
90th percentile
15.2%
Min
15.2%95.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.06 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 data95.8%
2o1high· Epoch AI Benchmarking Hub data91.6%
3Gemini 2.5 Pro· Epoch AI Benchmarking Hub data89.8%
4o3 Minihigh· Epoch AI Benchmarking Hub data89.6%
5Claude Sonnet 3.7· Epoch AI Benchmarking Hub data87.8%
6o3 Minimedium· Epoch AI Benchmarking Hub data86.3%
7QwQ 32B· Epoch AI Benchmarking Hub data83.5%
8DeepSeek R1· Epoch AI Benchmarking Hub data83.2%
9Gemini 2.0 Flash Experimentalthinking· Epoch AI Benchmarking Hub data78.2%
10O1 Minimedium· Epoch AI Benchmarking Hub data72.3%
11GPT-4.5 Preview· Epoch AI Benchmarking Hub data71.1%
12o3 Minilow· Epoch AI Benchmarking Hub data69.8%
13DeepSeek R1 Distill Llama 70b· Epoch AI Benchmarking Hub data67.6%
14DeepSeek V3· Epoch AI Benchmarking Hub data65.8%
15gemini-2.0-pro-exp-02-05· Epoch AI Benchmarking Hub data60.1%
16Gemini 2.0 Flash Experimental· Epoch AI Benchmarking Hub data59.1%
17QwQ 32B Preview· Epoch AI Benchmarking Hub data57.7%
18gemini-exp· Epoch AI Benchmarking Hub data57%
19Claude 3.5 Sonnet· Epoch AI Benchmarking Hub data56.7%
20Gemini 2.0 Flash· Epoch AI Benchmarking Hub data55.3%
21Grok 2· Epoch AI Benchmarking Hub data54.8%
22GPT 4o· Epoch AI Benchmarking Hub data53.9%
23DeepSeek R1 Distill Qwen 32B· Epoch AI Benchmarking Hub data52.3%
24Qwen2.5-Max· Epoch AI Benchmarking Hub data51.4%
25Llama 3.3 70B Instruct· Epoch AI Benchmarking Hub data50.8%
26Gemini 2.0 Flash-Lite Preview· Epoch AI Benchmarking Hub data50.1%
27Phi 4· Epoch AI Benchmarking Hub data47.8%
28Perplexity Sonar· Epoch AI Benchmarking Hub data46.3%
29Gemini 2.0 Flash Lite· Epoch AI Benchmarking Hub data44.9%
30Mistral Small2503· Epoch AI Benchmarking Hub data44.8%
31Gemma 3 27B IT· Epoch AI Benchmarking Hub data43.8%
32Mistral Large2411· Epoch AI Benchmarking Hub data43.5%
33Learnlm 1.5 Pro Experimental· Epoch AI Benchmarking Hub data43.4%
34Qwen 2.5 Coder 32B Instruct· Epoch AI Benchmarking Hub data42.1%
35Claude 3 Opus· Epoch AI Benchmarking Hub data40.6%
36Nova Litev1· Epoch AI Benchmarking Hub data36.7%
37Mistral Small· Epoch AI Benchmarking Hub data36.4%
38GPT-4o mini· Epoch AI Benchmarking Hub data32.8%
39Nova Prov1· Epoch AI Benchmarking Hub data32.6%
40Claude Haiku 3.5· Epoch AI Benchmarking Hub data28.1%
41Gemma 2 27B IT· Epoch AI Benchmarking Hub data28.1%
42Phi-3 Mini 4K Instruct· Epoch AI Benchmarking Hub data26.8%
43Nova Microv1· Epoch AI Benchmarking Hub data25.1%
44C4AI Command R 08 2024· Epoch AI Benchmarking Hub data21.9%
45Phi-3 Small 8K Instruct· Epoch AI Benchmarking Hub data15.9%
46Gemma 2 9B IT· Epoch AI Benchmarking Hub data15.2%

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