Skip to content

Benchmark

Arena ELO - Industry Entertainment and Sports and Media

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

Models measured
364
Scores recorded
421
Sources
1
IQ weight
1.17

Score distribution

1,279.8
Mean
1,314.7
Median
149.5
Std dev
1,058.5
10th percentile
1,441.1
90th percentile
784.8
Min
784.81,526.7
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.17 for chat models, measured over 364 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101DeepSeek V4 Flash· LMArena data1,399.8
102GPT-4.5 Preview2025-02-27· LMArena data1,399.3
103Nemotron 3 UltraNVFP4· LMArena data1,398.9
104MiMo V2.5· LMArena data1,398.9
105Beam 1.0 501B A23B· LMArena data1,398.8
106Qwen3.5 397B A17B· LMArena data1,398.5
107GPT-5.2 Chat20260210· LMArena data1,398.4
108Claude Opus 4.120250805· LMArena data1,398.2
109Inkling FP4· LMArena data1,397.7
110GPT 5.1· LMArena data1,397.2
111Kimi K2.5Instant· LMArena data1,396.8
112Claude Opus 4.1Thinking 16K, 20250805· LMArena data1,396.8
113nova-experimental-chat26-02-10· LMArena data1,396.7
114MiMo V2 Omni· LMArena data1,396.6
115DeepSeek V3.2· LMArena data1,395.2
116DeepSeek-V4-Flash PreviewHigh· LMArena data1,394.5
117MiniMax M3· LMArena data1,394.5
118DeepSeek V3.1Thinking· LMArena data1,393.6
119Gemini 2.5 Flash· LMArena data1,393.2
120LongCat-Flash-Chat Exp2602· LMArena data1,392.9
121DeepSeek V3.1 TerminusThinking· LMArena data1,391.7
122Grok 40709· LMArena data1,391
123Mistral Medium 3.12508· LMArena data1,389.9
124DeepSeek V3.2 ExpThinking· LMArena data1,389.8
125Qwen 3.8 27B· LMArena data1,389.4
126Hunyuan Vision 1.5 Thinking· LMArena data1,388.7
127Mistral Large 3· LMArena data1,386.8
128DeepSeek V3.2Thinking· LMArena data1,386.4
129Gemini 3.1 Flash Lite· LMArena data1,386.1
130Gemini 2.5 Flash Preview09-2025· LMArena data1,385.9
131DeepSeek V3.1 Terminus· LMArena data1,382.4
132Mistral Medium 3.5· LMArena data1,381.8
133Grok 4.1 FastReasoning· LMArena data1,381.5
134Grok 4 FastNon-Reasoning· LMArena data1,381
135Grok 4.7xHigh· LMArena data1,379.8
136GPT 5High· LMArena data1,379.2
137Qwen3 Max2025-09-23· LMArena data1,378.8
138DeepSeek V3.1· LMArena data1,378.4
139GPT 5.2· LMArena data1,377.7
140Kimi K2 Thinking· LMArena data1,377.6
141MiMo V2 FlashNon-Thinking· LMArena data1,375.6
142Qwen3 235B A22B Instruct 25072507· LMArena data1,374.2
143OpenAI o32025-04-16· LMArena data1,372.5
144GPT 5.2High· LMArena data1,372
145GPT-5.4 miniHigh· LMArena data1,371.8
146Grok 4.3· LMArena data1,370.6
147Qwen 3 235B A22B2507, thinking· LMArena data1,370
148Qwen3 VL 235Bthinking· LMArena data1,369.7
149Claude Haiku 4.520251001· LMArena data1,369.1
150Claude Opus 4Thinking 16K, 20250514· LMArena data1,368.8

Showing 101–150 of 421 results.

Data credits: LMArena, CC BY 4.0 (licence)