Skip to content

Benchmark

Arena ELO Text Factuality - Japanese

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

Models measured
47
Scores recorded
52
Sources
1
IQ weight
2.60

Score distribution

1,439
Mean
1,443.3
Median
42.4
Std dev
1,384.7
10th percentile
1,489.5
90th percentile
1,318.6
Min
1,318.61,522.9
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.60 for chat models, measured over 47 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 5High· LMArena data1,522.9
2Gemini 3.8 FlashHigh· LMArena data1,518.1
3GPT 5.5High· LMArena data1,515.2
4Kimi K3Max· LMArena data1,498.8
5GPT 5.5· LMArena data1,497
6GPT 5.4High· LMArena data1,489.5
7Claude Fable 5High· LMArena data1,487.6
8Gemini 3.1 Pro Preview· LMArena data1,484.8
9Claude Opus 4.6High· LMArena data1,481.5
10Claude Opus 4.7High· LMArena data1,481.3
11Claude Opus 4.6· LMArena data1,477.5
12GPT 5.4· LMArena data1,471.5
13Claude Opus 4.7· LMArena data1,467
14Gemini 3.5 FlashMedium· LMArena data1,465.5
15GPT-5.6 TerraxHigh· LMArena data1,463.5
16DeepSeek V4 Pro· LMArena data1,462.7
17Claude Sonnet 4.6· LMArena data1,460.4
18Claude Opus 4.8· LMArena data1,460
19GLM 5.2Max· LMArena data1,455.6
20Muse Spark 1.1· LMArena data1,453.2
21Inkling FP4· LMArena data1,452.1
22GLM 5.3 Flash· LMArena data1,451.3
23Kimi K2.6· LMArena data1,448.6
24Claude Opus 4.8High· LMArena data1,446.1
25Claude Sonnet 5High· LMArena data1,445.5
26GLM 5.1· LMArena data1,443.3
27Qwen3.5 397B A17B· LMArena data1,438.1
28DeepSeek-V4-Pro PreviewHigh· LMArena data1,435.5
29Grok 4.20 Multi-Agent0309, beta· LMArena data1,434.5
30Grok 4.20Reasoning, 0309, beta· LMArena data1,432.7
31Gemini 3 FlashMinimal· LMArena data1,428.8
32ERNIE 5.1· LMArena data1,426.1
33Kimi K2.5Thinking· LMArena data1,425
34Gemini 3.1 Flash Lite· LMArena data1,420.4
35GPT 5.2· LMArena data1,418.2
36MiMo V2.5 Pro· LMArena data1,418.1
37DeepSeek-V4-Flash PreviewHigh· LMArena data1,418
38Grok 4.3· LMArena data1,414.9
39GPT-6 LunaMax· LMArena data1,414.7
40Qwen3.7 Plus· LMArena data1,414.5
41DeepSeek V4 Flash· LMArena data1,413.5
42Mistral Large 3· LMArena data1,408
43GPT 5.4 NanoHigh· LMArena data1,396.1
44GPT-5.4 miniHigh· LMArena data1,395.7
45Qwen3.6 Plus· LMArena data1,391.5
46Doubao Seed 2.0 Pro· LMArena data1,390.4
47Qwen3.5 Flash· LMArena data1,384.7
48Claude Haiku 4.520251001· LMArena data1,379.6
49MiniMax M3· LMArena data1,378.4
50Step 3.5 Flash· LMArena data1,369.8

Showing 1–50 of 52 results.

Data credits: LMArena, CC BY 4.0 (licence)