Skip to content

Benchmark

Arena ELO Document Style-Controlled

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

Models measured
40
Scores recorded
46
Sources
1
IQ weight
2.29

Score distribution

1,462
Mean
1,466.7
Median
23.5
Std dev
1,432.6
10th percentile
1,494.4
90th percentile
1,405.6
Min
1,405.61,500.1
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.29 for chat models, measured over 40 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 Fable 5High· LMArena data1,500.1
2Claude Fable 5.1Max· LMArena data1,498.4
3Claude Opus 4.6High· LMArena data1,496.9
4Claude Opus 4.6· LMArena data1,495.6
5Claude Opus 4.7· LMArena data1,494.4
6Claude Opus 4.7High· LMArena data1,494.3
7GPT-5.6 SolxHigh· LMArena data1,486.5
8Claude Opus 5High· LMArena data1,485.9
9Claude Opus 4.8High· LMArena data1,485.4
10GPT 5.5High· LMArena data1,481.4
11Muse Spark 1.1· LMArena data1,479.8
12GPT 5.5· LMArena data1,479.2
13Claude Sonnet 4.6· LMArena data1,478.1
14Claude Opus 4.8· LMArena data1,476.1
15GPT-5.6 TerraxHigh· LMArena data1,474.2
16Muse Spark 1.3Max· LMArena data1,474.1
17GPT-6 AstraMax· LMArena data1,473.9
18GPT 5.4· LMArena data1,471.4
19Gemini 3.8 FlashHigh· LMArena data1,469.4
20Claude Sonnet 5High· LMArena data1,468.9
21Gemini 3.7 FlashHigh· LMArena data1,467.9
22Claude Opus 4.520251101· LMArena data1,467.5
23Muse Spark· LMArena data1,466.7
24Grok 4.5· LMArena data1,466.2
25GPT-5.6 LunaxHigh· LMArena data1,460.2
26GPT-5.5 Instant· LMArena data1,460.2
27Gemini 3.1 Pro Preview· LMArena data1,459
28Gemini 3.5 FlashMedium· LMArena data1,458.7
29Gemini 3.5 FlashHigh· LMArena data1,455.3
30Gemini 3.6 FlashHigh· LMArena data1,454.2
31Gemma 4 31B IT· LMArena data1,452
32Gemini 3 Pro Preview· LMArena data1,451
33Kimi K2.6· LMArena data1,450.4
34Grok 4.6High· LMArena data1,450.4
35Claude Sonnet 4.520250929· LMArena data1,448
36Qwen3.7 Plus· LMArena data1,446.4
37Grok 4.20Reasoning, 0309, beta· LMArena data1,439.9
38Gemini 3 Flash· LMArena data1,439.8
39Kimi K2.5Thinking· LMArena data1,437.6
40MiniMax M3· LMArena data1,435
41Gemini 2.5 Pro· LMArena data1,432.6
42GPT 5.2· LMArena data1,425.2
43GPT 5.2High· LMArena data1,421.7
44Claude Haiku 4.520251001· LMArena data1,418.7
45GPT 5.1· LMArena data1,416.3
46GLM 5V Turbo· LMArena data1,405.6

Data credits: LMArena, CC BY 4.0 (licence)