Skip to content

Benchmark

Arena ELO Text Factuality - English

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

Models measured
167
Scores recorded
189
Sources
1
IQ weight
2.15

Score distribution

1,448.4
Mean
1,453.2
Median
33.8
Std dev
1,399.6
10th percentile
1,487.4
90th percentile
1,336.9
Min
1,336.91,518.5
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.15 for chat models, measured over 167 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Grok 4.5· LMArena data1,472.2
52Muse Spark 1.1· LMArena data1,472.1
53Gemini 3 Flash· LMArena data1,471.3
54Claude Opus 4.520251101· LMArena data1,470.9
55GLM 5.3 Flash· LMArena data1,470.7
56GPT-5.6 LunaxHigh· LMArena data1,470.4
57Claude Opus 4.5High, 32K, 20251101· LMArena data1,470.4
58ERNIE 5.1· LMArena data1,469
59GPT 5.1High· LMArena data1,468.7
60DeepSeek V4 ProHigh, 20260813· LMArena data1,467.8
61Grok 4.20Reasoning, 0309, beta· LMArena data1,467.1
62Kimi K2.6· LMArena data1,465.8
63Grok 4.20 Multi-Agent0309, beta· LMArena data1,465.2
64GLM 5· LMArena data1,464.7
65DeepSeek V4 Pro· LMArena data1,464.7
66MiMo V2 Pro· LMArena data1,464.4
67Qwen3.7 Plus· LMArena data1,464.2
68GPT-6.1 TerraMax· LMArena data1,463.9
69Gemini 2.5 Pro· LMArena data1,463.9
70Claude Sonnet 4.520250929· LMArena data1,462.6
71Step 5 Preview· LMArena data1,462.5
72Grok 4.20 Beta 1· LMArena data1,462.4
73Claude Sonnet 4.5High, 32K, 20250929· LMArena data1,461.4
74DeepSeek-V4-Pro PreviewHigh· LMArena data1,461.2
75Claude Sonnet 5High· LMArena data1,461.1
76Gemma 4 31B IT· LMArena data1,460.9
77GPT-5.2 Chat20260210· LMArena data1,460.8
78Qwen3.6 Max Preview· LMArena data1,460.6
79GLM 4.6· LMArena data1,459.8
80GLM 4.7· LMArena data1,458.9
81Beam 1.0 501B A23B· LMArena data1,458.6
82GPT-6.2 AstraMax· LMArena data1,458.2
83Grok 4.6High· LMArena data1,458.2
84Hy3 (Deep Infra)· LMArena data1,457.2
85Inkling FP4· LMArena data1,456.7
86MiMo V2.5· LMArena data1,456.6
87DeepSeek V3.2 Exp· LMArena data1,455.8
88Kimi K2.5Thinking· LMArena data1,455.7
89Gemma 4 26B A4B· LMArena data1,455.7
90GPT-6.2 SolMax· LMArena data1,454.3
91Qwen3 Max Preview· LMArena data1,453.9
92ChatGPT-4o Latest20250326· LMArena data1,453.9
93ERNIE 5.00110· LMArena data1,453.6
94Gemini 3 FlashMinimal· LMArena data1,453.3
95Gemini 3.5 Flash Lite· LMArena data1,453.2
96LongCat-Flash-Chat Exp2602· LMArena data1,453.1
97MiMo V2 Omni· LMArena data1,451.9
98DeepSeek V3.2· LMArena data1,451.8
99Qwen3.5 397B A17B· LMArena data1,451.6
100Qwen3.6 Plus· LMArena data1,450.6

Showing 51–100 of 189 results.

Data credits: LMArena, CC BY 4.0 (licence)