Skip to content

Benchmark

Arena ELO Text Factuality - Exclude Ties

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

Models measured
169
Scores recorded
193
Sources
1
IQ weight
2.31

Score distribution

1,431.5
Mean
1,440
Median
52.6
Std dev
1,353.3
10th percentile
1,488
90th percentile
1,239.9
Min
1,239.91,545.9
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.31 for chat models, measured over 169 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Gemini 2.5 Pro· LMArena data1,466.5
52Claude Sonnet 4.6· LMArena data1,464.7
53GPT-6 LunaMax· LMArena data1,464.5
54ERNIE 5.1· LMArena data1,463.2
55Claude Opus 4.5High, 32K, 20251101· LMArena data1,462.8
56GLM 5.1· LMArena data1,461.7
57Claude Opus 4.520251101· LMArena data1,461.4
58GPT 5.1High· LMArena data1,460.8
59Kimi K2.6· LMArena data1,459.1
60Grok 4.20 Multi-Agent0309, beta· LMArena data1,458.8
61Grok 4.20Reasoning, 0309, beta· LMArena data1,458.8
62DeepSeek V4 Pro· LMArena data1,457.7
63GPT-5.6 LunaxHigh· LMArena data1,457.2
64nova-experimental-chat26-02-10· LMArena data1,456.8
65Qwen3.7 Plus· LMArena data1,455.4
66GLM 5· LMArena data1,454.4
67Grok 4.20 Beta 1· LMArena data1,453.2
68Inkling FP4· LMArena data1,452.8
69DeepSeek V4 ProHigh, 20260813· LMArena data1,452.5
70Qwen3.6 Max Preview· LMArena data1,452.5
71Claude Sonnet 5High· LMArena data1,451.8
72Gemini 3 FlashMinimal· LMArena data1,451.5
73Gemma 4 31B IT· LMArena data1,451.2
74GPT-6.1 TerraMax· LMArena data1,451
75MiMo V2.6 Flash· LMArena data1,451
76Grok 4.6High· LMArena data1,449.8
77ERNIE 5.00110· LMArena data1,449.3
78DeepSeek-V4-Pro PreviewHigh· LMArena data1,448.8
79Claude Sonnet 4.520250929· LMArena data1,448.6
80GPT-5.2 Chat20260210· LMArena data1,448.4
81DeepSeek V3.2 ExpThinking· LMArena data1,448.3
82GLM 4.6· LMArena data1,448.2
83Step 5 Preview· LMArena data1,446.8
84Hy3 (Deep Infra)· LMArena data1,446.7
85Kimi K2.5Thinking· LMArena data1,446.5
86Gemini 3.5 Flash Lite· LMArena data1,446.3
87Gemma 4 26B A4B· LMArena data1,445.8
88GLM 4.7· LMArena data1,445.1
89Qwen3.5 397B A17B· LMArena data1,444.7
90Claude Sonnet 4.5High, 32K, 20250929· LMArena data1,444.6
91ChatGPT-4o Latest20250326· LMArena data1,444.1
92ERNIE-5.0-Preview1203· LMArena data1,443.3
93MiMo V2 Pro· LMArena data1,442.3
94Qwen3 Max Preview· LMArena data1,442.3
95GPT-6.2 SolMax· LMArena data1,440.3
96GPT-5.4 miniHigh· LMArena data1,440.1
97GPT 5.1· LMArena data1,440
98GPT-5.5 Instant· LMArena data1,439.2
99DeepSeek V4 Flash· LMArena data1,438.4
100DeepSeek V3.2 Exp· LMArena data1,437.3

Showing 51–100 of 193 results.

Data credits: LMArena, CC BY 4.0 (licence)