Vision Arena

View rankings across multimodal, generative AI models capable of understanding and processing visual inputs

Last Updated

Jan 12, 2026

Total Votes

140,790

Total Models

/

Style Control

	Rank Spread
1	1◄─►2	gemini-3-pro	1309	±11	3,281	Google	Proprietary
2	1◄─►3	gemini-3-flash	1288	±14	1,973	Google	Proprietary
3	2◄─►10	gemini-3-flash (thinking-minimal)	1267	±22	641	Google	Proprietary
4	3◄─►8	gpt-5.1-high	1260	±13	2,112	OpenAI	Proprietary
5	3◄─►6	gemini-2.5-pro	1260	±7	20,792	Google	Proprietary
6	3◄─►10	gpt-5.1	1257	±13	2,302	OpenAI	Proprietary
7	4◄─►15	gpt-5-chat	1242	±8	11,375	OpenAI	Proprietary
8	5◄─►15	chatgpt-4o-latest-20250326	1239	±8	8,406	OpenAI	Proprietary
9	4◄─►20	gemini-2.5-flash-preview-09-2025	1237	±11	2,993	Google	Proprietary
10	7◄─►20	gpt-5-high	1234	±9	11,126	OpenAI	Proprietary
11	5◄─►23	ernie-5.0-preview-1220	1229	±15	1,632	Baidu	Proprietary
12	7◄─►22	qwen3-vl-235b-a22b-instruct	1227	±10	4,092	Alibaba	Apache 2.0
13	7◄─►20	o3-2025-04-16	1226	±8	15,218	OpenAI	Proprietary
14	9◄─►23	gpt-4.1-2025-04-14	1222	±8	12,156	OpenAI	Proprietary
15	7◄─►25	ernie-5.0-preview-1120	1222	±12	2,528	Baidu	Proprietary
16	9◄─►23	gemini-2.5-flash	1221	±7	14,685	Google	Proprietary
17	7◄─►28	claude-opus-4-20250514-thinking-16k	1216	±18	940	Anthropic	Proprietary
18	9◄─►29	qwen-vl-max-2025-08-13	1214	±17	1,196	Alibaba	Proprietary
19	9◄─►31	claude-sonnet-4-20250514-thinking-32k	1212	±19	836	Anthropic	Proprietary
20	9◄─►31	claude-3-7-sonnet-20250219-thinking-32k	1212	±18	976	Anthropic	Proprietary
21	12◄─►27	o4-mini-2025-04-16	1210	±8	12,221	OpenAI	Proprietary
22	12◄─►28	gpt-5-mini-high	1207	±11	8,540	OpenAI	Proprietary
23	16◄─►31	gpt-4.1-mini-2025-04-14	1202	±9	11,209	OpenAI	Proprietary
24	13◄─►31	qwen3-vl-235b-a22b-thinking	1201	±15	1,464	Alibaba	Apache 2.0
25	16◄─►32	claude-opus-4-20250514	1196	±15	1,510	Anthropic	Proprietary
26	17◄─►36	claude-3-7-sonnet-20250219	1189	±18	979	Anthropic	Proprietary
27	21◄─►32	gemini-2.5-flash-lite-preview-06-17-thinking	1188	±8	11,266	Google	Proprietary
28	18◄─►36	hunyuan-vision-1.5-thinking	1187	±15	1,580	Tencent	Proprietary
29	17◄─►37	claude-sonnet-4-20250514	1186	±16	1,235	Anthropic	Proprietary
30	21◄─►36	gemini-2.5-flash-lite-preview-09-2025-no-thinking	1183	±11	3,002	Google	Proprietary
31	20◄─►42	claude-3-5-sonnet-20241022	1179	±18	1,063	Anthropic	Proprietary
32	25◄─►39	grok-4-0709	1176	±8	10,469	xAI	Proprietary
33	27◄─►42	mistral-medium-2508	1169	±8	10,259	Mistral	Proprietary
34	27◄─►42	gemini-2.0-flash-001	1163	±11	3,403	Google	Proprietary
35	27◄─►44	gpt-5-nano-high	1161	±15	1,818	OpenAI	Proprietary
36	31◄─►43	gemma-3-27b-it	1160	±9	6,697	Google	Gemma
37	27◄─►45	glm-4.5v	1156	±16	1,286	Z.ai	MIT
38	32◄─►44	mistral-medium-2505	1156	±10	4,877	Mistral	Proprietary
39	31◄─►46	step-3	1153	±16	1,334	StepFun	Apache 2.0
40	30◄─►46	hunyuan-large-vision	1151	±19	886	Tencent	Proprietary
41	32◄─►45	llama-4-maverick-17b-128e-instruct	1151	±11	3,367	Meta	Llama 4
42	32◄─►46	step-1o-turbo-202506	1148	±16	1,365	StepFun	Proprietary
43	35◄─►46	mistral-small-2506	1140	±11	4,379	Mistral	Apache 2.0
44	36◄─►46	claude-3-5-haiku-20241022	1131	±19	996	Anthropic	Proprietary
45	38◄─►46	mistral-small-3.1-24b-instruct-2503	1131	±10	7,707	Mistral	Apache 2.0
46	40◄─►46	llama-4-scout-17b-16e-instruct	1126	±12	2,991	Meta	Llama

Vision Arena

Default Leaderboard Plots

Confidence Intervals on Model Strength (via Bootstrapping)

Battle Count for Each Combination of Models (without Ties)

Average Win Rate Against All Other Models (Uniform Sampling and No Ties)

Fraction of Model A Wins for All Non-tied A vs. B Battles