Skip to content

Benchmark

Arena ELO Style-Controlled

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

Models measured
366
Scores recorded
423
Sources
1
IQ weight
1.10

Score distribution

1,341.3
Mean
1,363.7
Median
121
Std dev
1,152.6
10th percentile
1,474.5
90th percentile
953.4
Min
953.41,527
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.10 for chat models, measured over 366 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Inkling FP4· LMArena data1,442.5
102Grok 4.3· LMArena data1,441.5
103Qwen3.5 397B A17B· LMArena data1,441.5
104GLM 4.7· LMArena data1,441.3
105DeepSeek-V4-Flash PreviewHigh· LMArena data1,439
106GPT 5.1· LMArena data1,438.5
107Gemma 4 26B A4B· LMArena data1,437.5
108GPT 5.2High· LMArena data1,437.3
109LongCat-Flash-Chat Exp2602· LMArena data1,435.8
110DeepSeek V4 Flash· LMArena data1,435.6
111GPT 5.2· LMArena data1,435.2
112GPT 5High· LMArena data1,435
113Qwen 3.8 27B· LMArena data1,434.6
114Qwen3 Max Preview· LMArena data1,434.4
115MiniMax M3· LMArena data1,434.4
116Beam 1.0 501B A23B· LMArena data1,434.3
117MiMo V2.5· LMArena data1,433.7
118Gemini 3.1 Flash Lite· LMArena data1,433.2
119OpenAI o32025-04-16· LMArena data1,431.5
120Kimi K2.5Instant· LMArena data1,430.7
121Grok 4.1 FastReasoning· LMArena data1,430.6
122MiMo V2 Omni· LMArena data1,430.4
123Kimi K2 Thinking· LMArena data1,430
124Nemotron 3 UltraNVFP4· LMArena data1,426.7
125nova-experimental-chat26-02-10· LMArena data1,426.7
126GPT-5 Chat· LMArena data1,426.5
127Mistral Medium 3.5· LMArena data1,426.3
128Claude Opus 4Thinking 16K, 20250514· LMArena data1,426.2
129Muse Glimmer· LMArena data1,425.1
130DeepSeek V3.2 ExpThinking· LMArena data1,425
131DeepSeek V3.2· LMArena data1,424.8
132GLM 4.6· LMArena data1,424.4
133Qwen3 Max2025-09-23· LMArena data1,423.2
134Mistral Large 4· LMArena data1,422.7
135DeepSeek V3.2Thinking· LMArena data1,422.6
136Qwen3 235B A22B Instruct 25072507· LMArena data1,422.2
137DeepSeek V3.2 Exp· LMArena data1,421.9
138DeepSeek R10528· LMArena data1,421.4
139Claude Haiku 5.5High· LMArena data1,420.8
140Grok 4 FastNon-Reasoning· LMArena data1,418.4
141Kimi K2 Preview0905· LMArena data1,418.3
142Kimi K2 Preview0711· LMArena data1,418.1
143ERNIE-5.0-Preview1022· LMArena data1,418
144DeepSeek V3.1· LMArena data1,417.1
145Qwen3.5 122B-A10B· LMArena data1,416.5
146DeepSeek V3.1 TerminusThinking· LMArena data1,416.2
147DeepSeek V3.1Thinking· LMArena data1,415.9
148NVIDIA Nemotron 3.5 Super VL 120BNVFP4· LMArena data1,415.6
149DeepSeek V3.1 Terminus· LMArena data1,414.7
150Mistral Large 3· LMArena data1,414.7

Showing 101–150 of 423 results.

Data credits: LMArena, CC BY 4.0 (licence)