AI Rankings
Best Vision AI Models: Ranked by Blind Votes on Image Questions
On Arena's vision board, people upload an image with a question, two anonymous models answer, and the voter picks the better answer. This page ranks models by the rating those votes produce. It covers understanding images, not making them.
Data updated
Top 20 at a glance
Tap a model to see its full page: scores, price and how it compares.
Bars start at 1270, not zero, so differences are easier to see.
Full ranking
All 85 models.Download the full table (CSV)Data credit: Arena (formerly LMArena), Epoch AI, LiteLLM model price list, used under open licences (details at the foot of the page).
| Rank and model | Arena vision rating | Likely range | Votes | Epoch index | Price per 1M tokens (in / out) |
|---|---|---|---|---|---|
1Claude Fable 5Anthropic | 1309 | 1302 to 1316 | 14,714 | 163.4 | $10.00 / $50.00 |
2Qwen 3.8 MaxAlibaba | 1300 | 1293 to 1308 | 9,957 | 157.4 | $2.00 / $6.00 |
3Claude Opus 4.6Anthropic | 1299 | 1293 to 1306 | 22,286 | 155.7 | $5.00 / $25.00 |
4Claude Opus 4.7Anthropic | 1299 | 1292 to 1305 | 23,139 | 156.8 | $5.00 / $25.00 |
5Gemini 3.7 FlashGoogle | 1295 | 1286 to 1304 | 5,369 | 158.0 | $0.75 / $3.75 |
6Muse SparkMeta | 1294 | 1285 to 1303 | 5,865 | 152.0 | Not measured |
7GPT-6.1 SolOpenAI | 1294 | 1281 to 1307 | 2,330 | 168.4 | $2.00 / $10.00 |
8Claude Opus 5.5Anthropic | 1293 | 1281 to 1305 | 3,035 | 168.8 | $4.00 / $20.00 |
9Muse Spark 1.2Meta | 1293 | 1278 to 1307 | 1,948 | 155.1 | $1.25 / $4.25 |
10Gemini 3.8 FlashGoogle | 1291 | 1281 to 1301 | 4,645 | 157.4 | $0.75 / $3.75 |
11Muse Spark 1.3Meta | 1290 | 1280 to 1299 | 5,392 | 157.2 | $1.25 / $4.25 |
12Gemini 3 ProGoogle | 1289 | 1282 to 1297 | 13,623 | 153.1 | $2.00 / $12.00 |
13Claude Opus 5Anthropic | 1288 | 1281 to 1295 | 17,743 | 163.9 | $5.00 / $25.00 |
14Claude Fable 5.1Anthropic | 1288 | 1278 to 1297 | 5,026 | 166.5 | $10.00 / $50.00 |
15GPT-5.6 SolOpenAI | 1286 | 1279 to 1294 | 11,439 | 162.8 | $4.00 / $20.00 |
16GPT-5.5OpenAI | 1286 | 1280 to 1292 | 28,781 | 159.7 | $5.00 / $30.00 |
17Claude Opus 4.8Anthropic | 1286 | 1279 to 1292 | 18,303 | 159.0 | $5.00 / $25.00 |
18GPT-6 AstraOpenAI | 1285 | 1275 to 1295 | 4,012 | 168.8 | $10.00 / $50.00 |
19GPT-5.4OpenAI | 1284 | 1278 to 1290 | 30,747 | 157.3 | $2.50 / $15.00 |
20Gemini 3.5 FlashGoogle | 1282 | 1276 to 1289 | 14,061 | 154.9 | $1.50 / $9.00 |
21Gemini 3.6 FlashGoogle | 1282 | 1274 to 1290 | 8,979 | 154.5 | $0.75 / $3.75 |
22Muse Spark 1.1Meta | 1280 | 1273 to 1287 | 12,804 | 154.1 | $1.25 / $4.25 |
23Gemini 3.1 ProGoogle | 1280 | 1274 to 1285 | 47,114 | 155.3 | $2.00 / $12.00 |
24Grok 4.5xAI | 1279 | 1272 to 1286 | 12,599 | 154.1 | $2.00 / $6.00 |
25GPT-5.2 ChatOpenAI | 1278 | 1271 to 1284 | 16,465 | Not measured | $1.75 / $14.00 |
26GPT-5.5 InstantOpenAI | 1276 | 1267 to 1285 | 7,590 | 142.4 | Not measured |
27Claude Sonnet 4.6Anthropic | 1275 | 1269 to 1281 | 27,408 | 152.8 | $3.00 / $15.00 |
28Gemini 3 FlashGoogle | 1273 | 1267 to 1278 | 39,131 | 151.9 | $0.50 / $3.00 |
29GLM-5.3-FlashZ.ai ยท Open weights | 1271 | 1265 to 1278 | 16,254 | 153.1 | $0.15 / $0.50 |
30GPT-5.6 TerraOpenAI | 1268 | 1261 to 1275 | 11,842 | 160.1 | $2.00 / $12.00 |
31Gemini 3.5 Flash-LiteGoogle | 1267 | 1259 to 1275 | 9,238 | 144.9 | $0.30 / $2.50 |
32GPT-6 SolOpenAI | 1266 | 1255 to 1277 | 3,221 | 163.7 | $2.00 / $10.00 |
33Claude Sonnet 5.5Anthropic | 1265 | 1252 to 1278 | 2,564 | 166.2 | $2.00 / $10.00 |
34Kimi K2.6Moonshot AI ยท Open weights | 1265 | 1258 to 1271 | 16,368 | 150.9 | $0.95 / $4.00 |
35Qwen3.7-PlusAlibaba | 1263 | 1257 to 1270 | 14,212 | 147.0 | $0.36 / $1.44 |
36Grok 4.6xAI | 1262 | 1254 to 1271 | 7,261 | 156.6 | $2.00 / $6.00 |
37Claude Sonnet 5Anthropic | 1262 | 1255 to 1269 | 14,034 | 156.9 | $2.00 / $10.00 |
38DeepSeek V4.1 FlashDeepSeek ยท Open weights | 1261 | 1251 to 1272 | 4,100 | 155.9 | $0.30 / $1.20 |
39GPT-5.6 LunaOpenAI | 1257 | 1249 to 1264 | 11,055 | 156.5 | $0.20 / $1.20 |
| 1255 | 1249 to 1261 | 26,796 | Not measured | $1.25 / $2.50 | |
41Kimi K2.5Moonshot AI ยท Open weights | 1252 | 1247 to 1258 | 28,249 | 148.1 | $0.60 / $3.00 |
42Mistral Large 4Mistral AI | 1252 | 1233 to 1271 | 1,008 | 150.0 | $0.68 / $2.09 |
43GPT-5.1OpenAI | 1251 | 1243 to 1259 | 9,848 | 149.8 | $1.25 / $10.00 |
44GPT-5.4 MiniOpenAI | 1250 | 1244 to 1257 | 25,734 | 148.7 | $0.75 / $4.50 |
| 1250 | 1244 to 1256 | 24,532 | Not measured | $1.25 / $2.50 | |
46Grok 4.7xAI | 1250 | 1239 to 1260 | 3,736 | 154.4 | $2.00 / $6.00 |
47Gemini 2.5 Pro (Jun 2025)Google | 1248 | 1243 to 1252 | 88,755 | 145.1 | $1.25 / $10.00 |
48Qwen3.5 397B-A17BAlibaba ยท Open weights | 1246 | 1241 to 1251 | 32,657 | 146.8 | $0.60 / $3.60 |
49MiMo-V2.6-ProXiaomi ยท Open weights | 1245 | 1230 to 1259 | 1,754 | Not measured | $0.43 / $0.87 |
50GPT-5.2OpenAI | 1243 | 1237 to 1249 | 19,022 | 153.7 | $1.75 / $14.00 |
51Qwen 3.8 27BAlibaba ยท Open weights | 1242 | 1234 to 1251 | 7,614 | 151.7 | $0.45 / $3.00 |
52Gemma 4 26B A4BGoogle ยท Open weights | 1242 | 1235 to 1248 | 23,922 | 141.9 | $0.15 / $0.60 |
53GPT-6 LunaOpenAI | 1240 | 1230 to 1251 | 3,662 | 156.8 | $0.10 / $0.50 |
54Gemini 3.1 Flash-LiteGoogle | 1235 | 1229 to 1240 | 27,657 | 144.3 | $0.25 / $1.50 |
55MiniMax-M3MiniMax ยท Open weights | 1233 | 1227 to 1240 | 19,844 | 146.9 | $0.30 / $1.20 |
56Gemini 2.5 Flash (Sep 2025)Google | 1230 | 1220 to 1240 | 4,792 | 142.9 | $0.30 / $2.50 |
57GPT-4.5OpenAI | 1225 | 1213 to 1237 | 2,925 | 136.7 | $75.00 / $150 |
58o3OpenAI | 1216 | 1209 to 1222 | 45,567 | 146.7 | $2.00 / $8.00 |
59GPT-4.1OpenAI | 1214 | 1207 to 1221 | 40,998 | 136.7 | $2.00 / $8.00 |
60GPT-5OpenAI | 1211 | 1204 to 1218 | 34,445 | 150.0 | $1.25 / $10.00 |
61Inkling-SmallThinking Machines ยท Open weights | 1209 | 1201 to 1217 | 8,074 | 150.3 | $0.47 / $1.20 |
62Claude Opus 4Anthropic | 1208 | 1191 to 1224 | 1,290 | 142.5 | $15.00 / $75.00 |
63Claude Sonnet 4Anthropic | 1207 | 1190 to 1224 | 1,144 | 141.6 | $3.00 / $15.00 |
64o4-miniOpenAI | 1201 | 1194 to 1208 | 41,316 | 145.5 | $1.10 / $4.40 |
65GPT-4.1 miniOpenAI | 1201 | 1193 to 1208 | 40,201 | 134.9 | $0.40 / $1.60 |
66Mistral Medium 3.5Mistral AI ยท Open weights | 1200 | 1190 to 1209 | 5,375 | 141.4 | $1.50 / $7.50 |
67GPT-5.4 NanoOpenAI | 1199 | 1193 to 1205 | 25,624 | 146.0 | $0.20 / $1.25 |
68Claude 3.7 SonnetAnthropic | 1193 | 1177 to 1208 | 1,436 | 141.0 | $3.00 / $15.00 |
68o1OpenAI | 1193 | 1182 to 1203 | 3,694 | 141.7 | $15.00 / $60.00 |
70GPT-5 miniOpenAI | 1184 | 1176 to 1192 | 28,943 | 145.3 | $0.25 / $2.00 |
71Grok 4xAI | 1184 | 1177 to 1192 | 30,173 | 146.2 | $3.00 / $15.00 |
72GPT-4o (May 2024)OpenAI | 1162 | 1153 to 1171 | 23,273 | 129.0 | $5.00 / $15.00 |
73Claude 3.5 Sonnet (October 2024)Anthropic | 1161 | 1153 to 1169 | 10,374 | 133.4 | $3.00 / $15.00 |
74Gemma 3 27BGoogle ยท Open weights | 1160 | 1152 to 1168 | 16,982 | 130.0 | $0.12 / $0.30 |
75Mistral Medium 3Mistral AI | 1155 | 1147 to 1164 | 10,793 | 134.0 | $1.50 / $7.50 |
76GPT-5 nanoOpenAI | 1147 | 1136 to 1159 | 3,928 | 139.2 | $0.05 / $0.40 |
77Claude 3.5 SonnetAnthropic | 1145 | 1135 to 1155 | 21,624 | 130.0 | $3.00 / $15.00 |
78Claude 3.5 HaikuAnthropic | 1127 | 1111 to 1144 | 1,398 | 127.1 | $0.90 / $4.50 |
79GPT-4o (Aug 2024)OpenAI | 1118 | 1106 to 1130 | 3,376 | 128.8 | $2.50 / $10.00 |
80GPT-4 Turbo (Apr 2024)OpenAI | 1112 | 1100 to 1124 | 13,391 | 127.3 | $10.00 / $30.00 |
81GPT-4o miniOpenAI | 1097 | 1089 to 1105 | 17,341 | 126.5 | $0.15 / $0.60 |
82GPT-4.1 nanoOpenAI | 1089 | 1070 to 1107 | 1,211 | 129.5 | $0.10 / $0.40 |
83Claude 3 OpusAnthropic | 1062 | 1051 to 1072 | 15,565 | 126.9 | $15.00 / $75.00 |
84Claude 3 SonnetAnthropic | 1016 | 1005 to 1028 | 12,314 | 120.8 | $3.00 / $15.00 |
85Claude 3 HaikuAnthropic | 1000 | 987 to 1012 | 13,380 | 118.8 | $0.25 / $1.25 |
How to read this leaderboard
Higher is better, and only the gaps between models matter. This board uses Arena's style-controlled ratings. Likely ranges that overlap mean the models are effectively tied. Only models that accept images appear here.
Questions people ask
What kind of questions are asked?
Voters attach an image and ask about it, such as describing a photo, reading a chart or explaining a diagram. The two anonymous answers are compared and the voter picks one.
Does this rank image generators?
No. This board is about models that read and understand images. Image-making tools are a different kind of test.
Does the top model suit my work best?
Not always. A leaderboard compares models on one kind of test. Price, speed, privacy rules, the apps you already use and the type of work you do all matter. Use the rank as a starting point, then try two or three models on your own tasks.
Why is a model I know missing?
A model appears only when an open source has published results for it. New models can take days or weeks to be measured, and models with no published score are left out rather than guessed.
Where does the data come from?
Ratings come from the open leaderboard dataset published by Arena (arena.ai), licensed CC BY 4.0. We keep the latest published ratings, match models to our other sources and rank them. Arena does not endorse this site.
How often is this leaderboard updated?
We check our open data sources every day and show the date of the data at the top of the page. A model only moves when its source publishes new results, so a number can stay the same for weeks.
More leaderboards
Different questions need different tests. Pick the one that matches your work.
- Overall intelligenceWhich AI model is the smartest overall?Leads: Claude Opus 5.5 (168.8)270 models ranked
- Human preferenceWhich AI do people prefer in blind tests?Leads: Gemini 4 Argon (1517)166 models ranked
- CodingWhich AI model is best for coding?Leads: Gemini 4 Argon (1552)166 models ranked
- Web developmentWhich AI model builds the best websites and apps?Leads: Claude Opus 5.5 (1807)85 models ranked
- Open-weights modelsWhich open-source AI model is best?Leads: Kimi K3 (157.9)123 models ranked
- Cheapest strong modelsWhat is the cheapest AI model that is still good?Cheapest: gpt-oss-20b ($0.10)119 models ranked
- Best valueWhich AI models give the most intelligence for the money?Line starts at: gpt-oss-20b ($0.10)7 models ranked
- Context windowWhich AI model can read the most text at once?Leads: GPT-5.4 (1.1M tokens)193 models ranked
- GPQA DiamondWhich AI model scores highest on GPQA Diamond?Leads: Gemini 4 Argon (96.4%)183 models ranked
- Humanity's Last ExamWhich AI model scores highest on Humanity's Last Exam?Leads: GPT-6 Astra (54.8%)41 models ranked
- SWE-bench VerifiedWhich AI model scores highest on SWE-bench Verified?Leads: Claude Opus 4.7 (83.5%)32 models ranked
- Aider PolyglotWhich AI model scores highest on Aider Polyglot?Leads: GPT-5 (88.0%)46 models ranked
- Terminal-BenchWhich AI model scores highest on Terminal-Bench?Leads: GPT-5.3 Codex (64.7%)25 models ranked
- FrontierMath (Tiers 1-3)Which AI model scores highest on FrontierMath (Tiers 1-3)?Leads: GPT-6 Astra (93.7%)87 models ranked
- ARC-AGI-2Which AI model scores highest on ARC-AGI-2?Leads: GPT-6 Astra (95.0%)86 models ranked
- SimpleQA VerifiedWhich AI model scores highest on SimpleQA Verified?Leads: Gemini 4 Argon (76.0%)82 models ranked
- SimpleBenchWhich AI model scores highest on SimpleBench?Leads: Claude Fable 5 (81.9%)80 models ranked
- DeepSWEWhich AI model scores highest on DeepSWE?Leads: GPT-6 Astra (74.1%)26 models ranked
Where this data comes from
- Arena (formerly LMArena): Human-preference ratings: Arena (arena.ai) leaderboard dataset, CC BY 4.0. We keep the latest published ratings, match them to models from other sources and rank them. Ratings are not changed. Arena does not endorse this site.
- Epoch AI: Benchmark data: Epoch AI, 'Capabilities & benchmarking', epoch.ai/benchmarks (CC BY 4.0). We match models across sources, rank them and show them in charts. Scores are not changed. Epoch AI does not endorse this site.
- LiteLLM model price list: Prices and context windows: LiteLLM model price list (github.com/BerriAI/litellm, MIT licence). We use the model maker's own price entry when there is one. Prices change often, so always confirm on the maker's site before you buy.
Newest data on this page: 26 Oct 2026. Prices are list prices and change often, so confirm on the maker's site before you buy. Sponsors never change a score or an order. How we calculate this ยท Report a mistake