Skip to content

Benchmark

LiveBench story_generation

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
1.56

Score distribution

71%
Mean
71.2%
Median
5.9%
Std dev
62.9%
10th percentile
77.5%
90th percentile
56.9%
Min
56.9%82.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.56 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 3.8 Flashhigh· LiveBench data82.8%
2Gemini 3.7 Flashhigh· LiveBench data82.3%
3Qwen 3.8 Flash Next· LiveBench data81.4%
4GPT 5.4xhigh· LiveBench data79%
5Grok 4.6· LiveBench data78.3%
6Muse Spark 1.3xhigh· LiveBench data77.5%
7Gemini 3.1 Pro Preview· LiveBench data77.3%
8Claude Sonnet 5.5xhigh· LiveBench data77.1%
9Qwen3.8 Max· LiveBench data76.5%
10Qwen3.7 Max· LiveBench data76.3%
11Gemini 3.6 Flashhigh· LiveBench data75.6%
12Kimi K3· LiveBench data75.4%
13Gemini 3.5 Flashhigh· LiveBench data75.2%
14DeepSeek V4 Pro 0813· LiveBench data74.9%
15Grok 4.7xhigh· LiveBench data74.9%
16GPT-5.2 Codex· LiveBench data74.8%
17Inkling FP4xhigh· LiveBench data74.2%
18DeepSeek V4 Flash Vision Exp· LiveBench data73.8%
19Muse Spark 1.2xhigh· LiveBench data73.7%
20Claude Haiku 5.5xhigh· LiveBench data73.6%
21Grok 4.5· LiveBench data73.6%
22Qwen 3.8 27B· LiveBench data73.1%
23GPT 5.5xhigh· LiveBench data72.5%
24Grok Build 0.1· LiveBench data72.5%
25GLM 5.3· LiveBench data72.1%
26Claude Opus 5.5xhigh· LiveBench data71.2%
27Grok 4.3· LiveBench data70.6%
28GPT-6.1 Solxhigh· LiveBench data70.6%
29DeepSeek V4 Flash 0731· LiveBench data70.5%
30Claude Opus 4.7xhigh· LiveBench data69.4%
31Nemotron 3 Ultra· LiveBench data69.2%
32Muse Spark 1.1xhigh· LiveBench data69.2%
33Claude Sonnet 4.6auto, medium· LiveBench data68.9%
34Kimi K2.7 Code· LiveBench data68.6%
35DeepSeek V4 Pro· LiveBench data67.8%
36GPT 5.4 Nanoxhigh· LiveBench data67.7%
37Mistral Large 4high· LiveBench data67.5%
38Gemini 3.5 Flash Litehigh· LiveBench data67.5%
39DeepSeek V4 Flash· LiveBench data67.3%
40Kimi K2.6thinking· LiveBench data66.9%
41GPT-5.4 minixhigh· LiveBench data66.6%
42Claude Opus 4.6auto, high· LiveBench data66.1%
43Claude Opus 4.564k, high· LiveBench data65.9%
44GPT 5.2high· LiveBench data65.7%
45MiniMax M3· LiveBench data63.6%
46GLM 5.2· LiveBench data62.9%
47Claude Sonnet 5xhigh· LiveBench data62.3%
48Qwen3.6 27B· LiveBench data60.1%
49GLM 5.3 Flash· LiveBench data59.9%
50Nemotron 3.5 Lightning 30B A3B· LiveBench data59.4%

Showing 1–50 of 51 results.