Skip to content

Benchmark

Arena ELO Text Factuality - Creative Writing

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

Models measured
142
Scores recorded
160
Sources
1
IQ weight
2.07

Score distribution

1,418.6
Mean
1,420.8
Median
43.1
Std dev
1,361.3
10th percentile
1,468.6
90th percentile
1,295.9
Min
1,295.91,508.7
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.07 for chat models, measured over 142 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 Opus 5.5High· LMArena data1,508.7
2Claude Fable 5.2Max· LMArena data1,504.2
3Gemini 4 CarbonHigh· LMArena data1,503.5
4Gemini 4 ArgonHigh· LMArena data1,500.3
5Gemini 4 NeonHigh· LMArena data1,499.6
6Claude Fable 5.1Max· LMArena data1,499.1
7Claude Opus 4.6High· LMArena data1,489.8
8Gemini 4 HydrogenHigh· LMArena data1,485.3
9Claude Opus 5Max· LMArena data1,478.3
10Claude Fable 5High· LMArena data1,478.2
11Gemini 3 Pro Preview· LMArena data1,478.1
12Muse WonderMax· LMArena data1,477
13Claude Opus 4.7High· LMArena data1,473.4
14Qwen3.8 Max· LMArena data1,471.4
15Gemini 3.7 FlashHigh· LMArena data1,471.3
16Claude Opus 4.6· LMArena data1,469.4
17Gemini 3.8 FlashHigh· LMArena data1,468.6
18Claude Opus 5High· LMArena data1,468.3
19Claude Opus 4.7· LMArena data1,468
20Gemini 3.5 FlashHigh· LMArena data1,466.9
21Gemini 3.1 Pro Preview· LMArena data1,465.7
22Muse Spark 1.3Max· LMArena data1,463.5
23MiMo V2.6 Pro· LMArena data1,462.6
24Gemini 3 Flash· LMArena data1,462.2
25Qwen3.5-Max-Preview· LMArena data1,462
26Gemini 3.5 FlashMedium· LMArena data1,461.1
27Claude Sonnet 5.5xHigh· LMArena data1,458.1
28Gemini 3.6 FlashHigh· LMArena data1,457.1
29GLM 5.3Max· LMArena data1,456.3
30GPT 5.5High· LMArena data1,455.9
31Claude Opus 4.8High· LMArena data1,455.7
32GPT-5.6 SolxHigh· LMArena data1,454
33GPT 5.5· LMArena data1,453.9
34GLM 5.2Max· LMArena data1,450.7
35GLM 5.1· LMArena data1,449.3
36Claude Opus 4.8· LMArena data1,449.1
37Gemini 2.5 Pro· LMArena data1,448.2
38Claude Opus 4.520251101· LMArena data1,448.2
39GPT 5.4High· LMArena data1,448
40Kimi K3Max· LMArena data1,448
41Claude Sonnet 4.520250929· LMArena data1,447.7
42Grok 4.5· LMArena data1,446.7
43DeepSeek V4 Pro· LMArena data1,445.9
44Claude Opus 4.5High, 32K, 20251101· LMArena data1,445.4
45ERNIE 5.1· LMArena data1,444.3
46Kimi K3.1 ProMax· LMArena data1,443.6
47Grok 4.20 Multi-Agent0309, beta· LMArena data1,443.1
48GLM 5· LMArena data1,442.7
49Grok 4.20 Beta 1· LMArena data1,442.2
50MiMo V2.5 Pro· LMArena data1,441.7

Showing 1–50 of 160 results.

Data credits: LMArena, CC BY 4.0 (licence)