Skip to content

Benchmark

LiveBench math_comp

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
1.90

Score distribution

95.3%
Mean
96.1%
Median
2%
Std dev
93.1%
10th percentile
97.1%
90th percentile
86.3%
Min
86.3%98%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.90 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
1Claude Opus 4.7xhigh· LiveBench data98%
2Claude Haiku 5.5xhigh· LiveBench data97.1%
3Claude Opus 5.5xhigh· LiveBench data97.1%
4Claude Sonnet 5.5xhigh· LiveBench data97.1%
5DeepSeek V4 Flash· LiveBench data97.1%
6DeepSeek V4 Flash Vision Exp· LiveBench data97.1%
7DeepSeek V4 Pro 0813· LiveBench data97.1%
8GPT-6.1 Solxhigh· LiveBench data97.1%
9Grok 4.5· LiveBench data97.1%
10Grok 4.6· LiveBench data97.1%
11Grok 4.7xhigh· LiveBench data97.1%
12Inkling FP4xhigh· LiveBench data97.1%
13Mistral Large 4high· LiveBench data97.1%
14Muse Spark 1.2xhigh· LiveBench data97.1%
15Muse Spark 1.3xhigh· LiveBench data97.1%
16Qwen3.7 Max· LiveBench data97.1%
17DeepSeek V4 Flash 0731· LiveBench data96.1%
18DeepSeek V4 Pro· LiveBench data96.1%
19Gemini 3.1 Pro Preview· LiveBench data96.1%
20Gemini 3.7 Flashhigh· LiveBench data96.1%
21GLM 5.2· LiveBench data96.1%
22GLM 5.3· LiveBench data96.1%
23GPT 5.5xhigh· LiveBench data96.1%
24Kimi K2.6thinking· LiveBench data96.1%
25Muse Spark 1.1xhigh· LiveBench data96.1%
26Nemotron 3 Ultra· LiveBench data96.1%
27Claude Opus 4.564k, high· LiveBench data95.1%
28Claude Opus 4.6auto, high· LiveBench data95.1%
29Claude Sonnet 5xhigh· LiveBench data95.1%
30Gemini 3.5 Flashhigh· LiveBench data95.1%
31Gemini 3.8 Flashhigh· LiveBench data95.1%
32GPT 5.2high· LiveBench data95.1%
33GPT 5.4 Nanoxhigh· LiveBench data95.1%
34GPT-5.2 Codex· LiveBench data95.1%
35Grok 4.3· LiveBench data95.1%
36Grok Build 0.1· LiveBench data95.1%
37Kimi K3· LiveBench data95.1%
38Qwen3.8 Max· LiveBench data95.1%
39Claude Sonnet 4.6auto, medium· LiveBench data94.1%
40Gemini 3.6 Flashhigh· LiveBench data94.1%
41GPT 5.4xhigh· LiveBench data94.1%
42GPT-5.4 minixhigh· LiveBench data94.1%
43Kimi K2.7 Code· LiveBench data94.1%
44Qwen 3.8 27B· LiveBench data94.1%
45Qwen 3.8 Flash Next· LiveBench data94.1%
46GLM 5.3 Flash· LiveBench data93.1%
47Qwen3.6 Plus· LiveBench data93.1%
48Gemini 3.5 Flash Litehigh· LiveBench data92.2%
49Qwen3.6 27B· LiveBench data92.2%
50MiniMax M3· LiveBench data91.2%

Showing 1–50 of 51 results.