Skip to content

Benchmark

Arena ELO Text Factuality - Chinese

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

Models measured
108
Scores recorded
123
Sources
1
IQ weight
2.06

Score distribution

1,492.7
Mean
1,492
Median
35.3
Std dev
1,451.2
10th percentile
1,534.8
90th percentile
1,389.6
Min
1,389.61,562.4
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.06 for chat models, measured over 108 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Step 3.5 Flash· LMArena data1,464.7
102Solar Pro 4· LMArena data1,462.9
103GPT-5.3 Chat· LMArena data1,462.5
104GPT-6 SolMax· LMArena data1,458.7
105ChatGPT-4o Latest20250326· LMArena data1,458.5
106Mistral Large 3· LMArena data1,456.3
107MiniMax M2.7· LMArena data1,455.2
108Mistral Medium 3.12508· LMArena data1,454.6
109Claude Opus 4.120250805· LMArena data1,453.4
110nova-experimental-chat11-10· LMArena data1,452.5
111Qwen3.5 Flash· LMArena data1,451.2
112Grok 4.1Thinking· LMArena data1,449.2
113Claude Haiku 4.520251001· LMArena data1,448.3
114Claude Opus 4.1Thinking 16K, 20250805· LMArena data1,447.9
115Grok 4.3· LMArena data1,447.1
116Grok 4.1 FastReasoning· LMArena data1,443.8
117Gemini 2.5 Flash-Lite PreviewNon-Thinking, 09-2025· LMArena data1,441.4
118GPT 5.4 NanoHigh· LMArena data1,426.2
119MiniMax M2.5· LMArena data1,416.6
120Trinity Large Thinking· LMArena data1,402.8
121Nemotron 3.5 Lightning 30B A3BNVFP4· LMArena data1,398.5
122Nemotron 3 Nano 30B A3BBF16· LMArena data1,394.5
123Trinity Large Preview· LMArena data1,389.6

Showing 101–123 of 123 results.

Data credits: LMArena, CC BY 4.0 (licence)