Skip to content

Benchmark

LiveBench typos

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
2.17

Score distribution

78%
Mean
80%
Median
8.5%
Std dev
70%
10th percentile
86%
90th percentile
54%
Min
54%90%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.17 for chat models, measured over 51 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 3.7 Flashhigh· LiveBench data90%
2Gemini 3.8 Flashhigh· LiveBench data90%
3Gemini 3.6 Flashhigh· LiveBench data88%
4GPT 5.5xhigh· LiveBench data88%
5Gemini 3.5 Flashhigh· LiveBench data86%
6Muse Spark 1.3xhigh· LiveBench data86%
7Qwen3.8 Max· LiveBench data86%
8Claude Opus 4.6auto, high· LiveBench data84%
9Claude Sonnet 5.5xhigh· LiveBench data84%
10DeepSeek V4 Flash Vision Exp· LiveBench data84%
11GLM 5.3· LiveBench data84%
12GPT-6.1 Solxhigh· LiveBench data84%
13Grok 4.5· LiveBench data84%
14Grok 4.6· LiveBench data84%
15Kimi K3· LiveBench data84%
16MiniMax M3· LiveBench data84%
17Qwen3.7 Max· LiveBench data84%
18DeepSeek V4 Flash 0731· LiveBench data82%
19DeepSeek V4 Pro 0813· LiveBench data82%
20Gemini 3.1 Pro Preview· LiveBench data82%
21GPT 5.2high· LiveBench data82%
22GPT 5.4xhigh· LiveBench data82%
23Qwen3.6 Plus· LiveBench data82%
24Claude Opus 4.7xhigh· LiveBench data80%
25Claude Opus 5.5xhigh· LiveBench data80%
26DeepSeek V4 Pro· LiveBench data80%
27Grok Build 0.1· LiveBench data80%
28Kimi K2.7 Code· LiveBench data80%
29Claude Opus 4.564k, high· LiveBench data78%
30GLM 5.3 Flash· LiveBench data78%
31Grok 4.3· LiveBench data78%
32Kimi K2.6thinking· LiveBench data78%
33Nemotron 3 Ultra· LiveBench data78%
34Qwen 3.8 Flash Next· LiveBench data78%
35Claude Sonnet 5xhigh· LiveBench data76%
36Inkling FP4xhigh· LiveBench data76%
37Qwen 3.8 27B· LiveBench data76%
38DeepSeek V4 Flash· LiveBench data74%
39Gemini 3.5 Flash Litehigh· LiveBench data74%
40GLM 5.2· LiveBench data74%
41Muse Spark 1.2xhigh· LiveBench data74%
42Claude Sonnet 4.6auto, medium· LiveBench data72%
43GPT-5.2 Codex· LiveBench data70%
44GPT-5.4 minixhigh· LiveBench data70%
45Grok 4.7xhigh· LiveBench data70%
46Qwen3.6 27B· LiveBench data70%
47Muse Spark 1.1xhigh· LiveBench data62%
48Mistral Large 4high· LiveBench data60%
49Nemotron 3.5 Lightning 30B A3B· LiveBench data58%
50Claude Haiku 5.5xhigh· LiveBench data54%

Showing 1–50 of 51 results.