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 Creative writingData Oct 19, 2026
Gemini 3 ProGoogle · 1342

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 · Creative writing

104 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 · Creative writing License
Score Rank Votes
1 Gemini 3 ProGoogle 1342±19 1 1,096 Proprietary
2 Claude Fable 5 (High)Anthropic 1338±19 2 1,159 Proprietary
3 Claude Opus 4.6 (High)Anthropic 1331±17 3 1,407 Proprietary
4 Claude Fable 5.1 (Max)Anthropic 1330±28 4 511 Proprietary
5 Muse SparkMeta 1327±33 5 346 Proprietary
6 Claude Opus 4.7 (High)Anthropic 1326±16 6 1,454 Proprietary
7 Gemini 3.1 Pro PreviewGoogle 1324±11 7 3,296 Proprietary
8 Claude Opus 4.7Anthropic 1321±17 8 1,397 Proprietary
9 Claude Opus 5.5 (High)Anthropic 1320±37 9 283 Proprietary
10 Claude Opus 4.6Anthropic 1319±15 10 1,733 Proprietary
11 Gemini 3 FlashGoogle 1314±12 11 2,968 Proprietary
12 Claude Opus 5 (High)Anthropic 1313±16 12 1,633 Proprietary
13 Qwen3.8-MaxAlibaba 1312±22 13 747 Proprietary
14 Gemini 3.7 Flash (High)Google 1312±28 14 505 Proprietary
15 GPT-5.5 (High)OpenAI 1301±14 15 2,038 Proprietary
16 Gemini 3.5 Flash (High)Google 1301±18 16 1,180 Proprietary
17 Gemini 3.5 Flash (Medium)Google 1301±19 17 1,127 Proprietary
18 Claude Opus 4.8 (High)Anthropic 1301±17 18 1,398 Proprietary
19 GPT-6.1 Sol (Max)OpenAI 1301±40 19 229 Proprietary
20 Gemini 3.8 Flash (High)Google 1299±29 20 459 Proprietary
21 Muse Spark 1.3 (Max)Meta 1296±28 21 493 Proprietary
22 Gemini 3 Flash (Minimal)Google 1293±13 22 2,497 Proprietary
23 Grok 4.5xAI 1293±19 23 1,094 Proprietary
24 Claude Opus 4.8Anthropic 1292±17 24 1,396 Proprietary
25 GPT-5.5OpenAI 1287±14 25 2,126 Proprietary
26 Muse Spark 1.3 (xHigh)Meta 1286±28 26 476 Proprietary
27 GPT-6 Astra (Max)OpenAI 1286±31 27 408 Proprietary
28 Gemma 4 31BGoogleOpen 1285±12 28 2,703 Apache 2.0
29 GPT-5.6 Sol (xHigh)OpenAI 1283±19 29 1,039 Proprietary
30 GPT-5.4 (High)OpenAI 1283±14 30 2,112 Proprietary
31 Grok 4.20 Multi-Agent Beta (0309)xAI 1281±15 31 1,692 Proprietary
32 Gemini 3.6 Flash (High)Google 1278±22 32 792 Proprietary
33 Grok 4.20 Beta (Reasoning, 0309)xAI 1276±15 33 1,706 Proprietary
34 DeepSeek-V4.1-Flash-20260910 (Max)DeepSeekOpen 1275±33 34 345 MIT
35 Kimi K2.6Moonshot AIOpen 1273±20 35 916 Modified MIT
36 Claude Sonnet 4.6Anthropic 1272±15 36 1,696 Proprietary
37 GPT-5.4OpenAI 1272±16 37 1,446 Proprietary
38 Gemini 2.5 Flash Preview (09-2025)Google 1270±33 38 318 Proprietary
39 GLM-5.3-FlashZ.aiOpen 1270±17 39 1,355 MIT
40 Kimi K2.5 (Thinking)Moonshot AIOpen 1269±14 40 1,768 Modified MIT
41 GPT-5.1OpenAI 1266±21 41 935 Proprietary
42 Gemini 2.5 ProGoogle 1265±12 42 3,682 Proprietary
43 Grok 4.6 (High)xAI 1262±24 43 714 Proprietary
44 Qwen3.5-397B-A17BAlibabaOpen 1260±13 44 2,304 Apache 2.0
45 Gemini 3.5 Flash-LiteGoogle 1260±21 45 862 Proprietary
46 GPT-5.5 InstantOpenAI 1259±32 46 361 Proprietary
47 Dola Seed 2.0 ProByteDance 1258±26 47 543 Proprietary
48 Qwen3.7-PlusAlibaba 1258±18 48 1,094 Proprietary
49 GPT-5.6 Luna (xHigh)OpenAI 1257±20 49 977 Proprietary
50 Gemma 4 26B A4BGoogleOpen 1256±16 50 1,495 Apache 2.0
51 ChatGPT-4o Latest (20250326)OpenAI 1254±18 51 1,245 Proprietary
52 Muse Spark 1.1Meta 1253±19 52 1,130 Proprietary
53 GLM-5V-TurboZ.ai 1252±12 53 2,924 Proprietary
54 GPT-5.6 Terra (xHigh)OpenAI 1248±20 54 992 Proprietary
55 GPT-5.1 (High)OpenAI 1248±22 55 795 Proprietary
56 Gemini 3.1 Flash-Lite PreviewGoogle 1248±14 56 1,883 Proprietary
57 Claude Sonnet 5 (High)Anthropic 1246±19 57 1,124 Proprietary
58 Claude Sonnet 5.5 (xHigh)Anthropic 1242±40 58 228 Proprietary
59 Gemini 2.5 FlashGoogle 1241±13 59 2,919 Proprietary
60 Qwen3.8-27BAlibabaOpen 1237±23 60 696 Apache 2.0
61 Kimi K2.5 (Instant)Moonshot AIOpen 1237±35 61 286 Modified MIT
62 Grok 4.3xAI 1237±14 62 2,011 Proprietary
63 GPT-5.2 Chat (20260210)OpenAI 1235±19 63 1,070 Proprietary
64 MiniMax-M3MiniMaxOpen 1234±17 64 1,339 MiniMax Community License
65 Grok 4.1 Fast (Reasoning)xAI 1233±21 65 877 Proprietary
66 ERNIE-5.0-Preview (1220)Baidu 1232±29 66 482 Proprietary
67 MiMo-V2.6-FlashXiaomiOpen 1231±36 67 303 MIT
68 NNVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)NVIDIAOpen 1231±44 68 193 OpenMDW-1.1
69 Grok 4 (0709)xAI 1230±20 69 1,227 Proprietary
70 MiMo-V2.5XiaomiOpen 1230±14 70 1,908 MIT
71 GPT-5.2OpenAI 1225±17 71 1,290 Proprietary
72 Qwen3.5-27BAlibabaOpen 1224±14 72 2,140 Apache 2.0
73 GPT-5.4 mini (High)OpenAI 1222±15 73 1,629 Proprietary
74 Qwen3.5-122B-A10BAlibabaOpen 1220±21 74 865 Apache 2.0
75 Qwen3-VL-235B-A22B-InstructAlibabaOpen 1220±20 75 974 Apache 2.0
76 GPT-6 Sol (Max)OpenAI 1219±38 76 286 Proprietary
77 Mistral Large 3Mistral AIOpen 1217±21 77 879 Apache 2.0
78 Grok 4.7 (xHigh)xAI 1217±33 78 370 Proprietary
79 GPT-5.2 (High)OpenAI 1213±17 79 1,330 Proprietary
80 GPT-5 ChatOpenAI 1210±19 80 1,297 Proprietary
81 GPT-4.1 (2025-04-14)OpenAI 1210±20 81 1,283 Proprietary
82 GPT-6 Luna (Max)OpenAI 1205±34 82 327 Proprietary
83 TInkling-SmallThinking MachinesOpen 1205±24 83 675 Apache 2.0
84 Qwen3-VL-235B-A22B-ThinkingAlibabaOpen 1203±44 84 178 Apache 2.0
85 o3 (2025-04-16)OpenAI 1201±18 85 1,584 Proprietary
86 Mistral Medium 3.1 (2508)Mistral AI 1200±17 86 1,550 Proprietary
87 Mistral Medium 3.5Mistral AIOpen 1198±33 87 343 Modified MIT
88 Gemini 2.5 Flash-Lite Preview (Non-Thinking, 09-2025)Google 1192±31 88 343 Proprietary
89 MiMo-V2-OmniXiaomi 1185±24 89 687 Proprietary
90 GPT-5 (High)OpenAI 1183±18 90 1,520 Proprietary
91 GPT-4.1 mini (2025-04-14)OpenAI 1182±21 91 1,192 Proprietary
92 GPT-5 mini (High)OpenAI 1181±21 92 1,017 Proprietary
93 Mistral Small 3.2 (2506)Mistral AIOpen 1181±29 93 466 Apache 2.0
94 Gemini 2.5 Flash-Lite Preview (Thinking, 06-17)Google 1179±21 94 1,188 Proprietary
95 Gemma 3 27B ITGoogleOpen 1177±25 95 627 Gemma
96 Gemini 2.0 Flash (001)Google 1172±39 96 250 Proprietary
97 o4-mini (2025-04-16)OpenAI 1166±20 97 1,258 Proprietary
98 GPT-5.4 nano (High)OpenAI 1159±17 98 1,531 Proprietary
99 Mistral Medium 3 (2505)Mistral AI 1146±28 99 480 Proprietary
100 Hunyuan Vision 1.5 ThinkingTencent 1146±43 100 209 Proprietary
101 Llama 4 Scout 17B 16E InstructMetaOpen 1126±41 101 238 Llama
102 Mistral Small 3.1 24B Instruct (2503)Mistral AIOpen 1123±24 102 771 Apache 2.0
103 GLM-4.6VZ.aiOpen 1113±41 103 241 MIT
104 Llama 4 Maverick 17B 128E InstructMetaOpen 1099±43 104 214 Llama 4

Scores: LMArena leaderboard-dataset (lmarena-ai on Hugging Face, CC-BY-4.0, vision config), Creative writing 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.