Best AI for Entity Recognition: Vision Arena

Vision Arena🔍Entity Recognition

View overall rankings across multimodal AI models capable of reasoning over visual inputs.

Oct 2, 2026

6,737 votes

50 models

	Rank Spread
1	122	gemini-3-pro Google · Proprietary	1291±36	248	$2 / $12	1M
2	122	gemini-3-flash Google · Proprietary	1288±30	365	$0.50 / $3	1M
3	122	gemini-3.1-pro-preview Google · Proprietary	1285±29	379	$1 / $6	1M
4	127	gemini-3-flash (thinking-minimal) Google · Proprietary	1273±31	320	$0.50 / $3	1M
5	129	gemini-3.1-flash-lite-preview Google · Proprietary	1269±34	257	$0.25 / $1.50	1M
6	132	gpt-5-high OpenAI · Proprietary	1261±33	432	$0.63 / $5	400K
7	132	gemini-2.5-pro Google · Proprietary	1250±22	916	$1.25 / $10	1M
8	143	grok-4.20-beta-0309-reasoning SpaceXAI · Proprietary	1241±43	165	$1.25 / $2.50	1M
9	144	claude-opus-4-6-high Anthropic · Proprietary	1239±47	121	$5 / $25	1M
10	144	gpt-5.2-chat-latest-20260210 OpenAI · Proprietary	1238±48	138	$1.75 / $14	128K
11	143	claude-opus-4-7-high Anthropic · Proprietary	1237±40	155	$5 / $25	1M
12	143	claude-opus-4-6 Anthropic · Proprietary	1237±40	183	$5 / $25	1M
13	143	grok-4-0709 SpaceXAI · Proprietary	1237±36	332	$1.25 / $2.50	256K
14	143	o3-2025-04-16 OpenAI · Proprietary	1232±31	495	$2 / $8	200K
15	145	gpt-5.1-high OpenAI · Proprietary	1231±49	113	$0.63 / $5	400K
16	143	kimi-k2.5-thinking Moonshot · Modified MIT	1229±33	270	$0.60 / $3	N/A
17	145	grok-4.20-multi-agent-beta-0309 SpaceXAI · Proprietary	1224±43	164	$1.25 / $2.50	1M
18	443	gemini-2.5-flash Google · Proprietary	1221±24	628	$0.30 / $2.50	1M
19	146	grok-4.3 SpaceXAI · Proprietary	1220±46	132	$1.25 / $2.50	1M
20	445	o4-mini-2025-04-16 OpenAI · Proprietary	1219±33	388	$1.10 / $4.40	200K
21	445	chatgpt-4o-latest-20250326 OpenAI · Proprietary	1218±32	295	$5 / $15	128K
22	148	grok-4-1-fast-reasoning SpaceXAI · Proprietary	1216±50	118	$0.20 / $0.50	128K
23	147	gpt-5.4 OpenAI · Proprietary	1215±47	134	$1.25 / $7.50	1.1M
24	150	gpt-5.5 OpenAI · Proprietary	1213±54	115	$5 / $30	1.1M
25	148	gpt-5.5-high OpenAI · Proprietary	1213±48	127	$5 / $30	1.1M
26	446	qwen3.5-397b-a17b Alibaba · Apache 2.0	1213±38	218	$0.55 / $3.50	262.1K
27	447	gpt-5-mini-high OpenAI · Proprietary	1211±40	282	$0.13 / $1	400K
28	550	gpt-5.2-high OpenAI · Proprietary	1200±42	168	$0.88 / $7	400K
29	648	gpt-4.1-mini-2025-04-14 OpenAI · Proprietary	1197±33	358	$0.40 / $1.60	1M
30	848	gemma-4-31b Google · Apache 2.0	1195±29	354	$0.99 / $1.49	131.1K
31	550	gpt-5.4-high OpenAI · Proprietary	1193±47	153	$2.50 / $15	1.1M
32	650	gpt-5.4-mini-high OpenAI · Proprietary	1192±41	182	$0.75 / $4.50	400K
33	850	gpt-5-chat OpenAI · Proprietary	1190±33	377	$5 / $30	N/A
34	650	gpt-5.1 OpenAI · Proprietary	1185±43	143	$0.63 / $5	400K
35	850	gpt-4.1-2025-04-14 OpenAI · Proprietary	1185±32	391	$2 / $8	1M
36	850	gemini-2.5-flash-lite-preview-06-17-thinking Google · Proprietary	1183±32	418	$0.10 / $0.40	1M
37	850	qwen3-vl-235b-a22b-instruct Alibaba · Apache 2.0	1181±43	151	$0.29 / $1.15	262.1K
38	850	claude-opus-4-7 Anthropic · Proprietary	1176±44	149	$5 / $25	1M
39	850	gpt-5.2 OpenAI · Proprietary	1173±41	176	$0.88 / $7	400K
40	850	glm-5v-turbo Z.ai · Proprietary	1171±40	224	$1.20 / $4	202.8K
41	850	gemma-3-27b-it Google · Gemma	1169±34	313	$0.08 / $0.45	131.1K
42	850	claude-sonnet-4-6 Anthropic · Proprietary	1169±46	158	$1.50 / $7.50	1M
43	850	mimo-v2.5 Xiaomi · MIT	1162±41	181	$0.14 / $0.28	1.1M
44	1550	gemma-4-26b-a4b Google · Apache 2.0	1156±36	233	N/A	N/A
45	1750	qwen3.5-27b Alibaba · Apache 2.0	1144±43	180	$0.26 / $2.60	262.1K
46	2150	mistral-small-2506 Mistral · Apache 2.0	1135±42	207	$0.10 / $0.30	32K
47	2350	mistral-small-3.1-24b-instruct-2503 Mistral · Apache 2.0	1130±41	240	$0.10 / $0.30	32K
48	2950	mistral-medium-2508 Mistral · Proprietary	1128±32	439	$0.40 / $2	131.1K
49	2550	mistral-medium-2505 Mistral · Proprietary	1125±40	200	$0.40 / $2	131.1K
50	2950	gpt-5.4-nano-high OpenAI · Proprietary	1119±42	175	$0.20 / $1.25	400K

Vision Arena🔍Entity Recognition

View overall rankings across multimodal AI models capable of reasoning over visual inputs.

Default Leaderboard Plots

Fraction of Model A Wins in A vs. B Battles (Ties Excluded)

Average Win Rate Against Each Opponent (Ties Excluded)

Battle Count for Each Pair of Models (Ties Excluded)

Confidence Intervals on Model Strength (via Bootstrapping)