Skip to content

Benchmark

Arena ELO Text Factuality - German

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

Models measured
49
Scores recorded
55
Sources
1
IQ weight
1.57

Score distribution

1,417.9
Mean
1,417.8
Median
29
Std dev
1,383.5
10th percentile
1,462
90th percentile
1,363.6
Min
1,363.61,475.3
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.57 for chat models, measured over 49 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 3 Pro Preview· LMArena data1,475.3
2GPT 5.5· LMArena data1,472
3Claude Opus 4.6High· LMArena data1,471.2
4GPT 5.4High· LMArena data1,466.4
5Claude Opus 5High· LMArena data1,462.8
6GPT 5.5High· LMArena data1,462
7Claude Opus 4.7High· LMArena data1,458.2
8Claude Opus 4.6· LMArena data1,456.8
9Gemini 3.1 Pro Preview· LMArena data1,452.4
10Claude Sonnet 5High· LMArena data1,447.9
11Claude Opus 4.8High· LMArena data1,447.9
12Gemini 2.5 Pro· LMArena data1,447.4
13DeepSeek V4 Pro· LMArena data1,442.3
14Gemini 3 FlashMinimal· LMArena data1,441.6
15Inkling FP4· LMArena data1,437.3
16Claude Opus 4.8· LMArena data1,436.4
17Grok 4.20 Multi-Agent0309, beta· LMArena data1,436.1
18Kimi K3Max· LMArena data1,434
19GLM 5.1· LMArena data1,427.8
20MiMo V2.5 Pro· LMArena data1,427.6
21Grok 4.1· LMArena data1,422.3
22Claude Opus 4.520251101· LMArena data1,421.2
23GPT 5.2· LMArena data1,420.6
24GPT 5.4· LMArena data1,420.6
25GLM 5.3 Flash· LMArena data1,420.4
26DeepSeek-V4-Pro PreviewHigh· LMArena data1,417.8
27MiniMax M3· LMArena data1,417.8
28GPT 5.1High· LMArena data1,417.8
29Qwen3.5 397B A17B· LMArena data1,416.3
30Qwen3 235B A22B Instruct 25072507· LMArena data1,414.5
31Kimi K2.5Thinking· LMArena data1,413.9
32Gemini 2.5 Flash· LMArena data1,410.6
33Grok 4.1Thinking· LMArena data1,406.5
34ChatGPT-4o Latest20250326· LMArena data1,406.5
35Grok 4.20Reasoning, 0309, beta· LMArena data1,404.8
36Claude Sonnet 4.520250929· LMArena data1,402.7
37Claude Sonnet 4.5High, 32K, 20250929· LMArena data1,401.9
38Claude Sonnet 4.6· LMArena data1,398.5
39Mistral Medium 3.12508· LMArena data1,397.9
40Mistral Large 3· LMArena data1,397.5
41GPT 5.4 NanoHigh· LMArena data1,395.9
42Gemini 3.1 Flash Lite· LMArena data1,395
43GPT-6 LunaMax· LMArena data1,391.3
44DeepSeek V3.2· LMArena data1,389.9
45Grok 4.3· LMArena data1,388
46Gemini 2.5 Flash-Lite PreviewNon-Thinking, 09-2025· LMArena data1,387.1
47MiMo V2 FlashNon-Thinking· LMArena data1,387.1
48Step 3.5 Flash· LMArena data1,386.6
49MiniMax M2.7· LMArena data1,383.9
50Grok 4.1 FastReasoning· LMArena data1,383.5

Showing 1–50 of 55 results.

Data credits: LMArena, CC BY 4.0 (licence)