Vision Arena

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

Feb 6, 2026

654,886 votes

94 models

/

Style Control

	Rank Spread
1	12	gemini-3-pro Google · Proprietary	1289±9	11,297
2	13	gemini-3-flash Google · Proprietary	1277±9	9,175
3	29	gpt-5.2-high OpenAI · Proprietary	1257±14	2,749
4	38	gemini-3-flash (thinking-minimal) Google · Proprietary	1256±10	7,313
5	39	gpt-5.1-high OpenAI · Proprietary	1252±10	7,299
6	310	kimi-k2.5-thinking Moonshot · Modified MIT	1251±13	2,979
7	310	gemini-2.5-pro Google · Proprietary	1246±6	79,747
8	614	chatgpt-4o-latest-20250326 OpenAI · Proprietary	1235±6	23,313
9	415	gpt-5.1 OpenAI · Proprietary	1235±9	7,974
10	322	kimi-k2.5-instant Moonshot · Modified MIT	1231±17	1,663
11	822	gemini-2.5-flash-preview-09-2025 Google · Proprietary	1225±10	5,293
12	822	gpt-4.5-preview-2025-02-27 OpenAI · Proprietary	1225±11	2,925
13	823	gpt-5.2 OpenAI · Proprietary	1223±14	3,013
14	822	gpt-5-chat OpenAI · Proprietary	1222±7	43,264
15	925	ernie-5.0-preview-1220 Baidu · Proprietary	1216±11	3,623
16	1023	o3-2025-04-16 OpenAI · Proprietary	1216±7	49,181
17	1025	gemini-2.5-flash Google · Proprietary	1213±6	48,047
18	1025	gpt-4.1-2025-04-14 OpenAI · Proprietary	1213±7	44,463
19	1027	qwen3-vl-235b-a22b-instruct Alibaba · Apache 2.0	1211±8	10,750
20	1027	gpt-5-high OpenAI · Proprietary	1208±8	37,581
21	1032	claude-opus-4-20250514-thinking-16k Anthropic · Proprietary	1206±15	1,495
22	1034	claude-sonnet-4-20250514-thinking-32k Anthropic · Proprietary	1205±16	1,361
23	1631	gpt-4.1-mini-2025-04-14 OpenAI · Proprietary	1201±8	43,674
24	1632	o4-mini-2025-04-16 OpenAI · Proprietary	1199±7	44,239
25	1437	claude-3-7-sonnet-20250219-thinking-32k Anthropic · Proprietary	1195±15	1,676
26	1937	o1-2024-12-17 OpenAI · Proprietary	1192±10	3,694
27	1937	claude-opus-4-20250514 Anthropic · Proprietary	1191±12	2,579
28	2137	gemini-2.5-flash-lite-preview-06-17-thinking Google · Proprietary	1188±8	39,110
29	2138	hunyuan-vision-1.5-thinking Tencent · Proprietary	1187±12	2,869
30	2139	qwen3-vl-235b-a22b-thinking Alibaba · Apache 2.0	1186±12	2,664
31	2139	claude-sonnet-4-20250514 Anthropic · Proprietary	1186±13	2,066
32	2439	grok-4-0709 xAI · Proprietary	1182±8	34,737
33	2439	gpt-5-mini-high OpenAI · Proprietary	1181±9	31,410
34	2240	qwen-vl-max-2025-08-13 Alibaba · Proprietary	1181±12	3,454
35	2539	gemini-1.5-pro-002 Google · Proprietary	1178±8	8,902
36	2541	claude-3-7-sonnet-20250219 Anthropic · Proprietary	1177±9	4,674
37	2545	gemini-2.5-flash-lite-preview-09-2025-no-thinking Google · Proprietary	1173±10	5,330
38	2946	gemini-2.0-flash-001 Google · Proprietary	1170±7	9,875
39	3551	gpt-4o-2024-05-13 OpenAI · Proprietary	1162±8	23,273
40	2953	glm-4.6v Z.ai · MIT	1161±14	2,611
41	3651	claude-3-5-sonnet-20241022 Anthropic · Proprietary	1161±7	10,568
42	3753	gemma-3-27b-it Google · Gemma	1156±8	18,534
43	3853	mistral-medium-2505 Mistral · Proprietary	1155±8	11,519
44	3755	glm-4.5v Z.ai · MIT	1154±12	3,576
45	3755	step-1o-turbo-202506 StepFun · Proprietary	1152±14	2,037
46	3756	hunyuan-large-vision Tencent · Proprietary	1151±16	1,440
47	3955	mistral-medium-2508 Mistral · Proprietary	1150±7	41,998
48	3955	claude-3-5-sonnet-20240620 Anthropic · Proprietary	1146±9	21,624
49	3955	llama-4-maverick-17b-128e-instruct Meta · Llama 4	1145±9	7,410
50	3958	gpt-5-nano-high OpenAI · Proprietary	1144±11	4,325
51	3958	step-3 StepFun · Apache 2.0	1144±12	3,558
52	4160	mistral-small-2506 Mistral · Apache 2.0	1139±9	11,713
53	4161	gemini-1.5-flash-002 Google · Proprietary	1139±9	7,241
54	4463	gemini-2.0-flash-lite-preview-02-05 Google · Proprietary	1133±10	3,991
55	4463	claude-3-5-haiku-20241022 Anthropic · Proprietary	1130±15	1,583
56	5063	mistral-small-3.1-24b-instruct-2503 Mistral · Apache 2.0	1126±9	30,955
57	5063	llama-4-scout-17b-16e-instruct Meta · Llama	1125±10	6,826
58	4963	step-1o-vision-32k-highres StepFun · Proprietary	1123±12	2,833
59	5263	qwen2.5-vl-72b-instruct Alibaba · Qwen	1121±10	3,768
60	5363	gpt-4o-2024-08-06 OpenAI · Proprietary	1118±12	3,376
61	5464	gemini-1.5-pro-001 Google · Proprietary	1117±11	16,734
62	5266	qwen2.5-vl-32b-instruct Alibaba · Apache 2.0	1116±15	1,490
63	5466	gpt-4-turbo-2024-04-09 OpenAI · Proprietary	1112±11	13,391
64	6268	gpt-4o-mini-2024-07-18 OpenAI · Proprietary	1097±7	17,347
65	6268	pixtral-large-2411 Mistral · MRL	1093±9	5,423
66	6172	gpt-4.1-nano-2025-04-14 OpenAI · Proprietary	1088±18	1,211
67	6470	qwen2-vl-72b Alibaba · Qwen	1085±9	5,937
68	6472	qwen-vl-max-1119 Alibaba · Proprietary	1084±16	1,422
69	6673	gemini-1.5-flash-8b-001 Google · Proprietary	1070±10	6,243
70	6774	claude-3-opus-20240229 Anthropic · Proprietary	1063±10	15,565
71	6674	step-1v-32k StepFun · Proprietary	1063±16	1,534
72	6774	gemini-1.5-flash-001 Google · Proprietary	1059±11	13,260
73	7078	molmo-72b-0924 AI2 · Apache 2.0	1047±13	3,048
74	6981	hunyuan-standard-vision-2024-12-31 Tencent · Proprietary	1043±21	809
75	7381	llama-3.2-vision-90b-instruct Meta · Llama 3.2	1032±8	8,682
76	7382	qwen2-vl-7b-instruct Aliaba · Apache 2.0	1031±10	5,766
77	7483	pixtral-12b-2409 Mistral · Apache 2.0	1025±9	7,511
78	7384	internvl2-26b OpenGVLab · MIT	1024±12	5,148
79	7386	amazon-nova-lite-v1.0 Amazon · Proprietary	1020±15	1,854
80	7486	amazon-nova-pro-v1.0 Amazon · Proprietary	1019±13	2,335
81	7485	claude-3-sonnet-20240229 Anthropic · Proprietary	1019±11	12,314
82	7788	yi-vision 01 AI · Proprietary	1003±18	1,219
83	7888	claude-3-haiku-20240307 Anthropic · Proprietary	1002±12	13,380
84	7690	c4ai-aya-vision-32b Cohere · CC-BY-NC-4.0	1000±22	847
85	7988	molmo-7b-d-0924 AI2 · Apache 2.0	996±13	2,815
86	8288	llama-3.2-vision-11b-instruct Meta · Llama 3.2	991±11	4,817
87	8092	nvila-internal-15b-v1 Nvidia · -	987±20	1,077
88	8292	llava-onevision-qwen2-72b-ov LLaVA · Apache 2.0	980±18	1,321
89	8792	llava-v1.6-34b LLaVA · Apache 2.0	966±12	4,531
90	8692	minicpm-v-2_6 OpenBMB · Apache 2.0	964±15	1,987
91	8692	cogvlm2-llama3-chat-19b Zhipu AI · CogVLM2	964±15	1,991
92	8792	internvl2-4b OpenGVLab · MIT	957±12	3,703
93	9393	phi-3.5-vision-instruct Microsoft · MIT	921±15	2,592
94	9494	phi-3-vision-128k-instruct Microsoft · MIT	883±18	1,401

Vision Arena

Default Leaderboard Plots

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

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)