Skip to content

Benchmark

Arena ELO Vision

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

Models measured
147
Scores recorded
168
Sources
1
IQ weight
1.77

Score distribution

1,185.4
Mean
1,222.2
Median
114.5
Std dev
999
10th percentile
1,304.2
90th percentile
811.6
Min
811.61,325
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.77 for chat models, measured over 147 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
101Claude Sonnet 4Thinking 32K, 20250514· LMArena data1,191.2
102Gemini 2.5 Flash-Lite PreviewThinking, 06-17· LMArena data1,187.3
103Step-1o Turbo202506· LMArena data1,185.5
104GPT-4.1 mini2025-04-14· LMArena data1,181.4
105Hunyuan Large Vision· LMArena data1,179.9
106Hunyuan Vision 1.5 Thinking· LMArena data1,179
107Step3· LMArena data1,175.9
108Claude Opus 420250514· LMArena data1,175.4
109Claude Sonnet 420250514· LMArena data1,175.4
110Mistral Medium 3.12508· LMArena data1,172.3
111Claude Sonnet 3.7Thinking 32K, 20250219· LMArena data1,168.7
112o12024-12-17· LMArena data1,168.3
113Gemma 3 27B IT· LMArena data1,164.4
114GLM 4.6v· LMArena data1,163.1
115Gemini 1.5 Pro002· LMArena data1,161.3
116GPT-5 nanoHigh· LMArena data1,159.1
117Gemini 2.0 Flash001· LMArena data1,157.8
118Mistral Medium 32505· LMArena data1,156.8
119GLM 4.5V· LMArena data1,154.3
120Qwen 2.5 VL 32B Instruct· LMArena data1,152.1
121Claude Sonnet 3.720250219· LMArena data1,150.8
122Mistral Small 3.22506· LMArena data1,142.1
123Llama 4 Maverick· LMArena data1,141.7
124Gemini 1.5 Flash002· LMArena data1,140.5
125GPT 4o2024-05-13· LMArena data1,136.5
126Mistral Small 3.1 24B Instruct2503· LMArena data1,136.2
127Claude 3.5 Sonnet20241022· LMArena data1,125.4
128Step-1o Vision 32KHigh Res· LMArena data1,121.7
129Claude 3.5 Sonnet20240620· LMArena data1,120.2
130Llama 4 Scout· LMArena data1,118.5
131Qwen2.5 VL 72B Instruct· LMArena data1,107.3
132Gemini 2.0 Flash-Lite Preview02-05· LMArena data1,100.3
133Claude Haiku 3.520241022· LMArena data1,092
134GPT-4 Turbo2024-04-09· LMArena data1,089.8
135Pixtral Large2411· LMArena data1,089.3
136Gemini 1.5 Pro001· LMArena data1,088.2
137Molmo 2 8B· LMArena data1,081
138GPT-4o mini2024-07-18· LMArena data1,065.5
139GPT 4o2024-08-06· LMArena data1,064.5
140GPT-4.1 nano2025-04-14· LMArena data1,063.1
141Qwen-VL Max1119· LMArena data1,057.5
142Qwen2-VL-72B· LMArena data1,046.8
143Step-1V 32K· LMArena data1,045.9
144Gemini 1.5 Flash-8B001· LMArena data1,044
145Claude 3 Opus20240229· LMArena data1,022.8
146Molmo 72B0924· LMArena data1,012.2
147Pixtral 12B2409· LMArena data1,008.5
148Gemini 1.5 Flash001· LMArena data1,007.9
149InternVL2 26B· LMArena data1,004.5
150Llama 3.2 90B Vision Instruct· LMArena data999.6

Showing 101–150 of 168 results.

Data credits: LMArena, CC BY 4.0 (licence)