Skip to content

Benchmark

LiveBench Overall

Scores are percentages.

Models measured
42
Scores recorded
46
Sources
1
IQ weight
2.06

Score distribution

53.5%
Mean
54.3%
Median
15%
Std dev
36.4%
10th percentile
75.7%
90th percentile
22.4%
Min
22.4%82.3%
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
1Gemini 2.5 Pro· Epoch AI Benchmarking Hub data82.3%
2GPT 5.1high· Epoch AI Benchmarking Hub data78.8%
3Claude Sonnet 3.7· Epoch AI Benchmarking Hub data76.1%
4o3 Minihigh· Epoch AI Benchmarking Hub data75.9%
5o1high· Epoch AI Benchmarking Hub data75.7%
6QwQ 32B· Epoch AI Benchmarking Hub data72%
7DeepSeek R1· Epoch AI Benchmarking Hub data71.6%
8o3 Minimedium· Epoch AI Benchmarking Hub data70%
9GPT-4.5 Preview· Epoch AI Benchmarking Hub data69%
10Gemini 2.0 Flash Experimentalthinking· Epoch AI Benchmarking Hub data66.9%
11DeepSeek V3· Epoch AI Benchmarking Hub data66.9%
12gemini-2.0-pro-exp-02-05· Epoch AI Benchmarking Hub data65.1%
13gemini-exp· Epoch AI Benchmarking Hub data64.1%
14o3 Minilow· Epoch AI Benchmarking Hub data62.5%
15Qwen2.5-Max· Epoch AI Benchmarking Hub data62.3%
16Gemini 2.0 Flash· Epoch AI Benchmarking Hub data61.5%
17Gemini 2.0 Flash Experimental· Epoch AI Benchmarking Hub data59.3%
18Claude 3.5 Sonnet· Epoch AI Benchmarking Hub data59%
19O1 Minimedium· Epoch AI Benchmarking Hub data57.8%
20GPT 4o· Epoch AI Benchmarking Hub data55.3%
21DeepSeek R1 Distill Llama 70b· Epoch AI Benchmarking Hub data54.5%
22Grok 2· Epoch AI Benchmarking Hub data54.3%
23Gemini 2.0 Flash Lite· Epoch AI Benchmarking Hub data54.3%
24Gemini 2.0 Flash-Lite Preview· Epoch AI Benchmarking Hub data53.2%
25Learnlm 1.5 Pro Experimental· Epoch AI Benchmarking Hub data52.2%
26Llama 3.3 70B Instruct· Epoch AI Benchmarking Hub data50.2%
27Gemma 3 27B IT· Epoch AI Benchmarking Hub data50%
28Claude 3 Opus· Epoch AI Benchmarking Hub data49.2%
29Mistral Large2411· Epoch AI Benchmarking Hub data48.4%
30Perplexity Sonar· Epoch AI Benchmarking Hub data46.9%
31Qwen 2.5 Coder 32B Instruct· Epoch AI Benchmarking Hub data46.2%
32DeepSeek R1 Distill Qwen 32B· Epoch AI Benchmarking Hub data45.5%
33Mistral Small2503· Epoch AI Benchmarking Hub data44%
34Nova Prov1· Epoch AI Benchmarking Hub data43.5%
35Claude Haiku 3.5· Epoch AI Benchmarking Hub data43.5%
36Mistral Small· Epoch AI Benchmarking Hub data42.6%
37Phi 4· Epoch AI Benchmarking Hub data41.6%
38GPT-4o mini· Epoch AI Benchmarking Hub data41.3%
39QwQ 32B Preview· Epoch AI Benchmarking Hub data40.3%
40Gemma 2 27B IT· Epoch AI Benchmarking Hub data38.2%
41Nova Litev1· Epoch AI Benchmarking Hub data36.4%
42Nova Microv1· Epoch AI Benchmarking Hub data29.6%
43Gemma 2 9B IT· Epoch AI Benchmarking Hub data28.7%
44C4AI Command R 08 2024· Epoch AI Benchmarking Hub data27.5%
45Phi-3 Small 8K Instruct· Epoch AI Benchmarking Hub data24%
46Phi-3 Mini 4K Instruct· Epoch AI Benchmarking Hub data22.4%

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