Skip to content

Benchmark

Arena ELO - Expert

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

Models measured
318
Scores recorded
371
Sources
1
IQ weight
1.30

Score distribution

1,333.9
Mean
1,371.6
Median
146.5
Std dev
1,113
10th percentile
1,491.8
90th percentile
953.7
Min
953.71,581.3
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.30 for chat models, measured over 318 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 3.1 Pro Preview· LMArena data1,483.6
52Claude Sonnet 4.5High, 32K, 20250929· LMArena data1,482.9
53GPT 5.4· LMArena data1,481
54Claude Opus 4.5High, 32K, 20251101· LMArena data1,480.7
55GPT-6.1 TerraMax· LMArena data1,478.7
56Qwen 3.8 27B· LMArena data1,478.6
57Step 5 Preview· LMArena data1,478.5
58Inkling 2 Small· LMArena data1,478.4
59MiMo V2 Pro· LMArena data1,478
60Muse Spark 1.1· LMArena data1,477.7
61GPT-5.6 LunaxHigh· LMArena data1,477.4
62GLM 5.1· LMArena data1,477.3
63Qwen3.6 Max Preview· LMArena data1,475.2
64Gemini 3 Pro Preview· LMArena data1,474.3
65Nemotron 3 UltraNVFP4· LMArena data1,473.3
66Claude Sonnet 4.520250929· LMArena data1,472.6
67GPT 5.1High· LMArena data1,469
68Inkling FP4· LMArena data1,468.6
69Grok 4.6High· LMArena data1,468.5
70Qwen3.7 Plus· LMArena data1,466.9
71Grok 4.5· LMArena data1,466.3
72Kimi K2.5Thinking· LMArena data1,465.5
73Beam 1.0 501B A23B· LMArena data1,465.1
74Gemma 4 31B IT· LMArena data1,464.8
75Gemini 3 Flash· LMArena data1,463.6
76DeepSeek V4 Pro· LMArena data1,463.1
77Qwen 3 235B A22B2507, thinking· LMArena data1,462.9
78Qwen3.5 397B A17B· LMArena data1,462.4
79MiniMax M3· LMArena data1,460.3
80MiMo V2.5· LMArena data1,459.3
81Hy3 (Deep Infra)· LMArena data1,459.1
82Muse Spark· LMArena data1,456.3
83DeepSeek V4 ProHigh, 20260813· LMArena data1,454.8
84Qwen3 Max Preview· LMArena data1,454.6
85GLM 5· LMArena data1,454.4
86Qwen3.6 Plus· LMArena data1,454.3
87LongCat-Flash-Chat Exp2602· LMArena data1,453.6
88Gemini 2.5 Pro· LMArena data1,452.5
89Mistral Large 4· LMArena data1,450.2
90DeepSeek-V4-Pro PreviewHigh· LMArena data1,448.6
91MiMo V2 Omni· LMArena data1,447.7
92Gemma 4 26B A4B· LMArena data1,447.7
93Grok 4.20 Multi-Agent0309, beta· LMArena data1,447.1
94MiniMax M2.7· LMArena data1,443.6
95GPT 5.2High· LMArena data1,443.5
96Claude Haiku 4.520251001· LMArena data1,443
97Qwen3 235B A22B Instruct 25072507· LMArena data1,441.9
98Qwen3 VL 235B· LMArena data1,441.9
99DeepSeek V4 Flash· LMArena data1,441.5
100LongCat Flash Chat· LMArena data1,441

Showing 51–100 of 371 results.

Data credits: LMArena, CC BY 4.0 (licence)