Skip to content

Benchmark

Arena ELO - Creative Writing

People compare model answers and choose which they prefer. Higher scores are better.

Models measured
364
Scores recorded
421
Sources
1
IQ weight
1.14

Score distribution

1,287.4
Mean
1,318.1
Median
144.2
Std dev
1,079.9
10th percentile
1,442.3
90th percentile
793.9
Min
793.91,532.7
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.14 for chat models, measured over 364 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 4 CarbonHigh· LMArena data1,532.7
2Gemini 4 ArgonHigh· LMArena data1,530.9
3Claude Opus 5.5High· LMArena data1,530.5
4Gemini 4 NeonHigh· LMArena data1,519
5Claude Fable 5.1Max· LMArena data1,515.9
6Claude Fable 5.2Max· LMArena data1,515.5
7Gemini 4 HydrogenHigh· LMArena data1,508.3
8Claude Opus 4.6High· LMArena data1,503.6
9Claude Fable 5High· LMArena data1,494.4
10Gemini 3.8 FlashHigh· LMArena data1,494.1
11Claude Opus 5Max· LMArena data1,493.8
12Muse WonderMax· LMArena data1,487.7
13Claude Opus 4.7High· LMArena data1,486.7
14Claude Opus 5High· LMArena data1,486.5
15Gemini 3.7 FlashHigh· LMArena data1,485.3
16Qwen3.8 Max· LMArena data1,483.7
17Claude Opus 4.6· LMArena data1,483.5
18Gemini 3.1 Pro Preview· LMArena data1,482.6
19Gemini 3 Pro Preview· LMArena data1,481.8
20Claude Opus 4.7· LMArena data1,479.7
21Gemini 3.5 FlashMedium· LMArena data1,470.7
22MiMo V2.6 Pro· LMArena data1,470.4
23Muse Spark 1.3Max· LMArena data1,469.3
24Gemini 3.5 FlashHigh· LMArena data1,469
25Gemini 3.6 FlashHigh· LMArena data1,466.5
26Qwen3.5-Max-Preview· LMArena data1,466.4
27Claude Sonnet 5.5xHigh· LMArena data1,463.3
28GLM 5.3Max· LMArena data1,460.3
29GLM 5.2Max· LMArena data1,459.6
30Muse Spark· LMArena data1,459.4
31Gemini 3 Flash· LMArena data1,456.3
32Claude Opus 4.8High· LMArena data1,455.2
33Gemini 2.5 Pro· LMArena data1,454.1
34GPT 5.5High· LMArena data1,453.5
35Kimi K3Max· LMArena data1,453.3
36GLM 5.1· LMArena data1,453.1
37GPT 5.5· LMArena data1,451.4
38GPT-5.6 SolxHigh· LMArena data1,450.7
39Qwen3.7 Max Preview· LMArena data1,447.8
40Claude Opus 4.8· LMArena data1,447
41DeepSeek V4 Pro· LMArena data1,446.3
42Claude Opus 4.5High, 32K, 20251101· LMArena data1,444.4
43Claude Sonnet 4.520250929· LMArena data1,442.3
44Claude Opus 4.520251101· LMArena data1,442.3
45Muse Spark 1.2xHigh· LMArena data1,441.8
46GLM 5.3 Flash· LMArena data1,441.7
47Grok 4.5· LMArena data1,441.6
48DeepSeek-V4-Pro PreviewHigh· LMArena data1,440.9
49ERNIE 5.1· LMArena data1,440.7
50Grok 4.20 Beta 1· LMArena data1,438.4

Showing 1–50 of 421 results.

Data credits: LMArena, CC BY 4.0 (licence)