Skip to content

Benchmark

Arena ELO Search Style-Controlled

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

Models measured
44
Scores recorded
45
Sources
1
IQ weight
2.33

Score distribution

1,190.7
Mean
1,196.1
Median
40.2
Std dev
1,138.2
10th percentile
1,228.6
90th percentile
1,061.9
Min
1,061.91,253.6
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.33 for chat models, measured over 42 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-5.6 SolxHigh· LMArena data1,253.6
2GPT-6.1 SolMax· LMArena data1,248.9
3GPT-6.2 AstraMax· LMArena data1,248.8
4GPT-6 AstraMax· LMArena data1,244.6
5GPT-6 SolMax· LMArena data1,228.6
6Claude Fable 5High· LMArena data1,228.3
7GPT-6.2 SolMax· LMArena data1,227.6
8Claude Fable 5.1High· LMArena data1,227.1
9GPT 5.5Search· LMArena data1,226.6
10Qwen3.8 Max· LMArena data1,226.2
11Claude Opus 4.6Search· LMArena data1,221.8
12Claude Opus 5.5Medium· LMArena data1,212.9
13Claude Opus 5High· LMArena data1,212.8
14Claude Opus 4.7· LMArena data1,209.8
15Gemini 3.1 Pro PreviewGrounding· LMArena data1,209.5
16GPT-6 LunaMax· LMArena data1,208.4
17Grok 4.20 Multi-Agent0309, beta· LMArena data1,207
18Grok 4.5· LMArena data1,205.6
19Gemini 3 Pro PreviewGrounding· LMArena data1,201
20Claude Sonnet 4.6Search· LMArena data1,201
21Grok 4.20 Beta 1· LMArena data1,199.7
22OpenAI o3Search· LMArena data1,198
23Grok 4.1 FastSearch· LMArena data1,196.1
24GPT 5.4Search· LMArena data1,195.3
25Claude Opus 4.8· LMArena data1,194
26Grok 4.3· LMArena data1,193.6
27ERNIE 5.1· LMArena data1,192.1
28Claude Sonnet 5Search· LMArena data1,190.5
29GPT 5Search· LMArena data1,189.6
30Gemini 3 FlashGrounding· LMArena data1,188
31Claude Opus 4.5Search· LMArena data1,186.3
32GPT 5.1Search· LMArena data1,183.7
33Amazon-Web-Search-Agent· LMArena data1,177.6
34Claude Sonnet 4.5Search· LMArena data1,174.3
35Claude Opus 4.1Search· LMArena data1,166.9
36Grok 4 FastSearch· LMArena data1,162.4
37GPT 5.2Search· LMArena data1,160.7
38Claude Opus 4Search· LMArena data1,151.5
39Gemini 2.5 ProGrounding· LMArena data1,145
40Sonar ProHigh· LMArena data1,142.3
41Sonar Reasoning ProHigh· LMArena data1,138.2
42GPT 5.2Search, Non-Reasoning· LMArena data1,135.9
43Grok 4Search· LMArena data1,123
44GPT 4oSearch· LMArena data1,083.8
45Diffbot-Small-XL· LMArena data1,061.9

Data credits: LMArena, CC BY 4.0 (licence)