Vision Model Leaderboard: Best Vision-Language Models

Ranked by LMArena Vision Arena scores: users upload an image, ask a question, and vote between two anonymous models’ answers.

Overall OCR Diagrams Homework Captioning Entity recognition Creative writing Humor English Chinese
#1 Entity recognitionData Oct 19, 2026
Gemini 3 ProGoogle · 1348

In LMArena Vision Arena data published Oct 19, 2026, Anthropic’s Claude Fable 5 (High) ranks #1 of 167 models with 1325 (14,714 votes).

Full leaderboard · Entity recognition

51 models
All Open weights

Sorted by LMArena score by default; ± is half the 95% confidence interval. Category columns show the model’s score on that sub-board; – means no score.

# Model LMArena · Entity recognition License
Score Rank Votes
1 Gemini 3 ProGoogle 1348±36 1 249 Proprietary
2 Gemini 3 FlashGoogle 1336±30 2 368 Proprietary
3 Gemini 3.1 Pro PreviewGoogle 1330±29 3 388 Proprietary
4 Gemini 3 Flash (Minimal)Google 1311±30 4 323 Proprietary
5 Gemini 3.1 Flash-Lite PreviewGoogle 1308±33 5 259 Proprietary
6 Grok 4 (0709)xAI 1288±35 6 332 Proprietary
7 Kimi K2.5 (Thinking)Moonshot AIOpen 1285±32 7 275 Modified MIT
8 Gemini 2.5 ProGoogle 1283±22 8 917 Proprietary
9 Claude Opus 4.7 (High)Anthropic 1278±41 9 154 Proprietary
10 Grok 4.20 Beta (Reasoning, 0309)xAI 1277±41 10 172 Proprietary
11 Claude Opus 4.6Anthropic 1274±38 11 189 Proprietary
12 Claude Opus 4.6 (High)Anthropic 1272±47 12 125 Proprietary
13 Grok 4.20 Multi-Agent Beta (0309)xAI 1265±43 13 169 Proprietary
14 Qwen3.5-397B-A17BAlibabaOpen 1265±37 14 222 Apache 2.0
15 GPT-5 (High)OpenAI 1263±30 15 432 Proprietary
16 Gemini 2.5 FlashGoogle 1252±24 16 628 Proprietary
17 Grok 4.1 Fast (Reasoning)xAI 1251±50 17 118 Proprietary
18 GPT-5.4OpenAI 1251±46 18 137 Proprietary
19 Qwen3-VL-235B-A22B-InstructAlibabaOpen 1249±42 19 151 Apache 2.0
20 GPT-5.2 Chat (20260210)OpenAI 1246±48 20 138 Proprietary
21 ChatGPT-4o Latest (20250326)OpenAI 1245±32 21 295 Proprietary
22 GLM-5V-TurboZ.ai 1241±38 22 231 Proprietary
23 o3 (2025-04-16)OpenAI 1234±30 23 494 Proprietary
24 GPT-5.2 (High)OpenAI 1232±41 24 168 Proprietary
25 GPT-5.1 (High)OpenAI 1232±49 25 112 Proprietary
26 GPT-5.5OpenAI 1231±52 26 119 Proprietary
27 GPT-5.4 (High)OpenAI 1229±45 27 159 Proprietary
28 Claude Opus 4.7Anthropic 1226±44 28 154 Proprietary
29 GPT-5 mini (High)OpenAI 1226±36 29 282 Proprietary
30 Gemma 4 31BGoogleOpen 1226±29 30 359 Apache 2.0
31 GPT-5.5 (High)OpenAI 1225±48 31 130 Proprietary
32 GPT-5 ChatOpenAI 1219±33 32 377 Proprietary
33 GPT-5.4 mini (High)OpenAI 1218±41 33 183 Proprietary
34 o4-mini (2025-04-16)OpenAI 1218±32 34 388 Proprietary
35 Grok 4.3xAI 1217±45 35 142 Proprietary
36 Qwen3.5-27BAlibabaOpen 1210±41 36 187 Apache 2.0
37 GPT-4.1 (2025-04-14)OpenAI 1208±32 37 392 Proprietary
38 GPT-5.2OpenAI 1197±41 38 176 Proprietary
39 MiniMax-M3MiniMaxOpen 1197±53 39 107 MiniMax Community License
40 Gemma 3 27B ITGoogleOpen 1196±34 40 313 Gemma
41 Claude Sonnet 4.6Anthropic 1195±46 41 162 Proprietary
42 Gemma 4 26B A4BGoogleOpen 1191±36 42 235 Apache 2.0
43 MiMo-V2.5XiaomiOpen 1189±40 43 185 MIT
44 GPT-4.1 mini (2025-04-14)OpenAI 1187±33 44 357 Proprietary
45 Gemini 2.5 Flash-Lite Preview (Thinking, 06-17)Google 1186±32 45 417 Proprietary
46 GPT-5.1OpenAI 1182±43 46 143 Proprietary
47 Mistral Small 3.1 24B Instruct (2503)Mistral AIOpen 1155±41 47 240 Apache 2.0
48 Mistral Small 3.2 (2506)Mistral AIOpen 1151±42 48 207 Apache 2.0
49 Mistral Medium 3.1 (2508)Mistral AI 1147±31 49 439 Proprietary
50 GPT-5.4 nano (High)OpenAI 1141±42 50 178 Proprietary
51 Mistral Medium 3 (2505)Mistral AI 1141±40 51 200 Proprietary

Scores: LMArena leaderboard-dataset (lmarena-ai on Hugging Face, CC-BY-4.0, vision config), Entity recognition published Oct 19, 2026. Please credit the source when citing.

Sources & methodology

  1. Scores are Bradley-Terry Arena ratings fitted from human votes between two anonymous answers to the same image prompt.
  2. Categories (OCR, diagrams, homework, captioning, Chinese/English…) are separate LMArena boards computed on prompts of that type, not a breakdown of the overall score; ranks can differ a lot.
  3. Fields without public data (price, context length, release date…) are not shown; blanks are –. Nothing is estimated.
  4. Synced daily. JSON API: /api/public/vision-leaderboard.php

FAQ

What is the best vision-language model right now?

In LMArena Vision Arena data published Oct 19, 2026, Anthropic’s Claude Fable 5 (High) ranks #1 of 167 models with 1325 (14,714 votes).

What is each vendor’s best vision model?

Top overall entry per vendor: Anthropic Claude Fable 5 (High) (1325, #1), Google Gemini 3.8 Flash (High) (1315, #7), Alibaba Qwen3.8-Max (1313, #10), Meta Muse Spark 1.3 (Max) (1308, #12), OpenAI GPT-5.4 (High) (1303, #18).

Which model is best at OCR?

On LMArena’s OCR board (123 models), Claude Fable 5 (High) is #1 with 1337.

What is the best open-weight vision model?

50 open-weight models are listed; the top one is GLM-5.3-Flash (1294, MIT).

How is this different from benchmarks like MMMU?

MMMU and similar benchmarks auto-score a fixed question set; this board comes from blind human votes on people’s own images and questions, reflecting real-use preference. They are not convertible.