Skip to content

Benchmark

LiveBench consecutive_events

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
1.91

Score distribution

70.9%
Mean
75.6%
Median
21.6%
Std dev
43.8%
10th percentile
90.2%
90th percentile
2.5%
Min
2.5%92.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.91 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 Sonnet 4.6auto, medium· LiveBench data92.7%
2GPT-6.1 Solxhigh· LiveBench data91.3%
3Claude Sonnet 5.5xhigh· LiveBench data90.9%
4Muse Spark 1.3xhigh· LiveBench data90.5%
5Claude Opus 5.5xhigh· LiveBench data90.4%
6DeepSeek V4 Pro 0813· LiveBench data90.2%
7DeepSeek V4 Flash Vision Exp· LiveBench data90%
8Kimi K3· LiveBench data89.8%
9Claude Opus 4.7xhigh· LiveBench data89.5%
10DeepSeek V4 Flash 0731· LiveBench data89.4%
11GPT-5.2 Codex· LiveBench data89%
12GPT 5.2high· LiveBench data88.9%
13GPT 5.5xhigh· LiveBench data88.8%
14Qwen3.8 Max· LiveBench data87.1%
15GPT 5.4xhigh· LiveBench data86.2%
16Grok 4.7xhigh· LiveBench data85.7%
17Gemini 3.1 Pro Preview· LiveBench data85.2%
18Qwen 3.8 27B· LiveBench data84.9%
19MiniMax M3· LiveBench data84.6%
20GLM 5.3 Flash· LiveBench data84.3%
21Muse Spark 1.2xhigh· LiveBench data82.5%
22Mistral Large 4high· LiveBench data80.5%
23Claude Opus 4.564k, high· LiveBench data79.4%
24GLM 5.2· LiveBench data79.3%
25Grok 4.6· LiveBench data78.1%
26Qwen 3.8 Flash Next· LiveBench data75.6%
27Grok 4.5· LiveBench data75.5%
28DeepSeek V4 Pro· LiveBench data75.4%
29Claude Sonnet 5xhigh· LiveBench data74.7%
30GLM 5.3· LiveBench data72.3%
31Qwen3.7 Max· LiveBench data71.8%
32Muse Spark 1.1xhigh· LiveBench data71.1%
33Claude Haiku 5.5xhigh· LiveBench data71%
34Qwen3.6 27B· LiveBench data70.4%
35Inkling FP4xhigh· LiveBench data69.9%
36Grok Build 0.1· LiveBench data69.7%
37Qwen3.6 Plus· LiveBench data67%
38Claude Opus 4.6auto, high· LiveBench data63.5%
39GPT-5.4 minixhigh· LiveBench data63%
40Gemini 3.7 Flashhigh· LiveBench data60.4%
41DeepSeek V4 Flash· LiveBench data59.9%
42GPT 5.4 Nanoxhigh· LiveBench data54.3%
43Kimi K2.6thinking· LiveBench data51.3%
44Kimi K2.7 Code· LiveBench data48.3%
45Gemini 3.5 Flashhigh· LiveBench data48%
46Gemini 3.6 Flashhigh· LiveBench data43.8%
47Grok 4.3· LiveBench data28%
48Gemini 3.5 Flash Litehigh· LiveBench data21.1%
49Gemini 3.8 Flashhigh· LiveBench data20.6%
50Nemotron 3 Ultra· LiveBench data18.8%

Showing 1–50 of 51 results.