Skip to content

Benchmark

LiveBench typescript

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
1.85

Score distribution

35.8%
Mean
36.7%
Median
12.8%
Std dev
20%
10th percentile
50%
90th percentile
6.7%
Min
6.7%60%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.85 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
1Muse Spark 1.3xhigh· LiveBench data60%
2DeepSeek V4 Flash Vision Exp· LiveBench data56.7%
3Qwen3.8 Max· LiveBench data56.7%
4Claude Opus 5.5xhigh· LiveBench data53.3%
5Kimi K3· LiveBench data53.3%
6Claude Sonnet 5xhigh· LiveBench data50%
7GLM 5.3· LiveBench data50%
8Muse Spark 1.2xhigh· LiveBench data50%
9Qwen 3.8 27B· LiveBench data50%
10Claude Sonnet 5.5xhigh· LiveBench data46.7%
11DeepSeek V4 Pro 0813· LiveBench data46.7%
12Gemini 3.7 Flashhigh· LiveBench data46.7%
13GLM 5.3 Flash· LiveBench data46.7%
14GPT-6.1 Solxhigh· LiveBench data46.7%
15Qwen 3.8 Flash Next· LiveBench data46.7%
16Claude Opus 4.7xhigh· LiveBench data43.3%
17GPT 5.4xhigh· LiveBench data43.3%
18GPT 5.5xhigh· LiveBench data43.3%
19Grok 4.6· LiveBench data43.3%
20Grok 4.7xhigh· LiveBench data43.3%
21Mistral Large 4high· LiveBench data43.3%
22Muse Spark 1.1xhigh· LiveBench data43.3%
23Claude Haiku 5.5xhigh· LiveBench data40%
24Gemini 3.8 Flashhigh· LiveBench data40%
25GLM 5.2· LiveBench data36.7%
26GPT 5.2high· LiveBench data36.7%
27Grok 4.5· LiveBench data36.7%
28Claude Opus 4.6auto, high· LiveBench data33.3%
29Gemini 3.5 Flashhigh· LiveBench data33.3%
30Grok Build 0.1· LiveBench data33.3%
31Inkling FP4xhigh· LiveBench data30%
32DeepSeek V4 Flash 0731· LiveBench data26.7%
33Gemini 3.5 Flash Litehigh· LiveBench data26.7%
34Gemini 3.6 Flashhigh· LiveBench data26.7%
35GPT 5.4 Nanoxhigh· LiveBench data26.7%
36Kimi K2.6thinking· LiveBench data26.7%
37Qwen3.7 Max· LiveBench data26.7%
38Claude Sonnet 4.6auto, medium· LiveBench data23.3%
39DeepSeek V4 Flash· LiveBench data23.3%
40DeepSeek V4 Pro· LiveBench data23.3%
41Gemini 3.1 Pro Preview· LiveBench data23.3%
42Kimi K2.7 Code· LiveBench data23.3%
43MiniMax M3· LiveBench data23.3%
44Qwen3.6 27B· LiveBench data23.3%
45Claude Opus 4.564k, high· LiveBench data20%
46GPT-5.2 Codex· LiveBench data20%
47GPT-5.4 minixhigh· LiveBench data20%
48Qwen3.6 Plus· LiveBench data20%
49Nemotron 3 Ultra· LiveBench data16.7%
50Grok 4.3· LiveBench data13.3%

Showing 1–50 of 51 results.