---
title: "Vision leaderboard"
description: "Rankings of multimodal models that reason over images: charts, screenshots, OCR and photos."
canonical: https://pirouter.ai/rankings/vision
last-updated: 2026-10-15
---

# Vision leaderboard

> Rankings of multimodal models that reason over images: charts, screenshots, OCR and photos.

- Updated: 2026-09-28
- 1,962,932 votes · 144 models
- Score · higher is better

## Full ranking

| Rank | Rank spread | Model | Vendor | Score | Votes |
|---|---|---|---|---|---|
| 1 | 1–7 | Claude Fable 5 (high) | Anthropic | 1310 ±7 | 13483 |
| 2 | 1–14 | Qwen3.8 (max) | Alibaba | 1301 ±8 | 10086 |
| 3 | 1–15 | Claude Opus 4.6 (high) | Anthropic | 1299 ±7 | 22350 |
| 4 | 1–16 | Claude Opus 4.7 | Anthropic | 1299 ±7 | 22948 |
| 5 | 1–25 | [Gemini 3.7 Flash \(high\)](https://pirouter.ai/models/google/gemini-3.7-flash) | Google | 1295 ±11 | 3425 |
| 6 | 1–25 | Muse Spark | Meta | 1294 ±9 | 5880 |
| 7 | 1–27 | Muse Spark 1.2 (xhigh) | Meta | 1292 ±14 | 1962 |
| 8 | 2–26 | Muse Spark 1.3 (max) | Meta | 1290 ±11 | 3472 |
| 9 | 2–25 | Gemini 3 Pro | Google | 1289 ±7 | 13624 |
| 10 | 2–25 | Claude Opus 5 (high) | Anthropic | 1289 ±7 | 15304 |
| 11 | 2–26 | [GPT-5.6 Sol \(xhigh\)](https://pirouter.ai/models/openai/gpt-5.6-sol) | OpenAI | 1287 ±8 | 10063 |
| 12 | 2–27 | [Claude Fable 5.1 \(max\)](https://pirouter.ai/models/anthropic/claude-fable-5.1) | Anthropic | 1287 ±10 | 4050 |
| 13 | 4–26 | [GPT-5.5](https://pirouter.ai/models/openai/gpt-5.5) | OpenAI | 1287 ±6 | 26870 |
| 14 | 3–26 | Claude Opus 4.8 (high) | Anthropic | 1287 ±7 | 18020 |
| 15 | 2–28 | [Gemini 3.8 Flash \(high\)](https://pirouter.ai/models/google/gemini-3.8-flash) | Google | 1286 ±12 | 2711 |
| 16 | 2–28 | [GPT-6 Astra \(max\)](https://pirouter.ai/models/openai/gpt-6-astra) | OpenAI | 1285 ±12 | 2869 |
| 17 | 5–27 | [Gemini 3.5 Flash \(high\)](https://pirouter.ai/models/google/gemini-3.5-flash) | Google | 1285 ±7 | 11983 |
| 18 | 5–27 | [GPT-5.4 \(high\)](https://pirouter.ai/models/openai/gpt-5.4) | OpenAI | 1284 ±6 | 28881 |
| 19 | 5–33 | [GLM 5.3 Flash](https://pirouter.ai/models/z-ai/glm-5.3-flash) | Z.ai | 1281 ±10 | 4832 |
| 20 | 5–33 | [Gemini 3.6 Flash \(high\)](https://pirouter.ai/models/google/gemini-3.6-flash) | Google | 1280 ±9 | 7076 |
| 21 | 5–28 | Gemini 3.1 Pro Preview | Google | 1280 ±5 | 45204 |
| 22 | 5–31 | Muse Spark 1.1 | Meta | 1280 ±7 | 10906 |
| 23 | 5–33 | Grok 4.5 | xAI | 1279 ±7 | 10681 |
| 24 | 5–32 | GPT-5.2 Chat Latest | OpenAI | 1278 ±7 | 16492 |
| 25 | 5–35 | GPT-5.5 Instant | OpenAI | 1276 ±9 | 7623 |
| 26 | 9–33 | Claude Sonnet 4.6 | Anthropic | 1275 ±6 | 27465 |
| 27 | 13–34 | Gemini 3 Flash | Google | 1273 ±5 | 39181 |
| 28 | 17–40 | [GPT-5.6 Terra \(xhigh\)](https://pirouter.ai/models/openai/gpt-5.6-terra) | OpenAI | 1268 ±8 | 9990 |
| 29 | 20–41 | Qwen3.7 Plus | Alibaba | 1266 ±7 | 12799 |
| 30 | 21–41 | Kimi K2.6 | Moonshot | 1265 ±7 | 16416 |
| 31 | 22–42 | [Claude Sonnet 5 \(high\)](https://pirouter.ai/models/anthropic/claude-sonnet-5) | Anthropic | 1265 ±7 | 12722 |
| 32 | 20–44 | [Gemini 3.5 Flash Lite](https://pirouter.ai/models/google/gemini-3.5-flash-lite) | Google | 1264 ±9 | 7200 |
| 33 | 20–44 | Grok 4.6 (high) | xAI | 1263 ±9 | 5374 |
| 34 | 26–42 | Gemma 4 31b | Google | 1263 ±6 | 38935 |
| 35 | 28–44 | Gemini 3 Flash | Google | 1260 ±5 | 37221 |
| 36 | 28–45 | [GPT-5.6 Luna \(xhigh\)](https://pirouter.ai/models/openai/gpt-5.6-luna) | OpenAI | 1259 ±8 | 10099 |
| 37 | 28–48 | Dola Seed 2.0 Pro | ByteDance | 1257 ±7 | 10986 |
| 38 | 28–48 | Grok 4.20 Beta 0309 Reasoning | xAI | 1256 ±6 | 26870 |
| 39 | 27–55 | MiMo V2.6 Flash | Xiaomi | 1253 ±15 | 1865 |
| 40 | 31–51 | Kimi K2.5 (thinking) | Moonshot | 1252 ±6 | 28367 |
| 41 | 29–52 | GPT-5.1 (high) | OpenAI | 1251 ±8 | 9843 |
| 42 | 33–51 | [GPT-5.4 Mini \(high\)](https://pirouter.ai/models/openai/gpt-5.4-mini) | OpenAI | 1251 ±6 | 25775 |
| 43 | 33–52 | Grok 4.20 Multi Agent Beta 0309 | xAI | 1250 ±6 | 24597 |
| 44 | 36–53 | Gemini 2.5 Pro | Google | 1247 ±5 | 88831 |
| 45 | 28–60 | MiMo V2.6 Pro | Xiaomi | 1246 ±15 | 1802 |
| 46 | 37–55 | Qwen3.5 397b A17b | Alibaba | 1245 ±5 | 31178 |
| 47 | 37–55 | Grok 4.3 | xAI | 1244 ±6 | 25944 |
| 48 | 39–57 | GPT-5.2 (high) | OpenAI | 1243 ±6 | 19051 |
| 49 | 37–58 | Qwen3.8 27b | Alibaba | 1242 ±9 | 6147 |
| 50 | 39–58 | Gemma 4 26b A4b | Google | 1242 ±7 | 23977 |
| 51 | 41–58 | Chatgpt 4o Latest | OpenAI | 1239 ±6 | 23429 |
| 52 | 39–61 | Kimi K2.5 Instant | Moonshot | 1237 ±11 | 4112 |
| 53 | 43–60 | Minimax M3 | MiniMax | 1237 ±7 | 17970 |
| 54 | 44–61 | Gemini 3.1 Flash Lite Preview | Google | 1235 ±6 | 27719 |
| 55 | 47–61 | MiMo V2.5 | Xiaomi | 1232 ±6 | 27858 |
| 56 | 44–68 | Gemini 2.5 Flash Preview 09 2025 | Google | 1230 ±10 | 4793 |
| 57 | 48–61 | GLM 5v Turbo | Z.ai | 1230 ±6 | 41294 |
| 58 | 51–69 | Qwen3.5 122b A10b | Alibaba | 1226 ±7 | 13259 |
| 59 | 51–69 | GPT-5 Chat | OpenAI | 1226 ±7 | 39918 |
| 60 | 47–71 | GPT-4.5 Preview 2025 02 27 | OpenAI | 1225 ±12 | 2925 |
| 61 | 53–76 | Ernie 5.0 Preview 1220 | Baidu | 1219 ±11 | 3351 |
| 62 | 57–71 | Qwen3.5 27b | Alibaba | 1217 ±6 | 28689 |
| 63 | 57–72 | MiMo V2 Omni | Xiaomi | 1216 ±7 | 10687 |
| 64 | 57–72 | O3 2025 04 16 | OpenAI | 1216 ±7 | 45606 |
| 65 | 58–71 | Gemini 2.5 Flash | Google | 1215 ±5 | 60141 |
| 66 | 57–75 | Qwen3 Vl 235b A22b Instruct | Alibaba | 1215 ±7 | 12734 |
| 67 | 57–76 | GPT-4.1 2025 04 14 | OpenAI | 1214 ±7 | 41027 |
| 68 | 60–79 | GPT-5 (high) | OpenAI | 1210 ±7 | 34464 |
| 69 | 57–83 | Claude Opus 4 (thinking) | Anthropic | 1207 ±16 | 1290 |
| 70 | 60–80 | Inkling Small | Thinky | 1207 ±9 | 6658 |
| 71 | 57–84 | Claude Sonnet 4 (thinking) | Anthropic | 1206 ±17 | 1145 |
| 72 | 65–82 | GPT-4.1 Mini 2025 04 14 | OpenAI | 1201 ±8 | 40235 |
| 73 | 66–82 | O4 Mini 2025 04 16 | OpenAI | 1201 ±7 | 41354 |
| 74 | 65–82 | Mistral Large 3 | Mistral | 1200 ±8 | 7880 |
| 75 | 63–84 | Mistral Medium 3.5 | Mistral | 1200 ±10 | 5391 |
| 76 | 68–82 | [GPT-5.4 Nano \(high\)](https://pirouter.ai/models/openai/gpt-5.4-nano) | OpenAI | 1200 ±6 | 25685 |
| 77 | 68–84 | Grok 4.1 Fast Reasoning | xAI | 1196 ±7 | 13401 |
| 78 | 65–87 | Claude 3.7 Sonnet (thinking) | Anthropic | 1193 ±16 | 1438 |
| 79 | 68–86 | O1 2024 12 17 | OpenAI | 1193 ±11 | 3694 |
| 80 | 70–86 | Gemini 2.5 Flash Lite Preview 06 17 (thinking) | Google | 1188 ±7 | 35953 |
| 81 | 69–87 | Qwen3 Vl 235b A22b (thinking) | Alibaba | 1188 ±12 | 2389 |
| 82 | 70–88 | Qwen Vl Max 2025 08 13 | Alibaba | 1185 ±12 | 3173 |
| 83 | 74–87 | Grok 4 0709 | xAI | 1184 ±7 | 30195 |
| 84 | 75–88 | GPT-5 Mini (high) | OpenAI | 1183 ±8 | 28971 |
| 85 | 78–89 | Gemini 1.5 Pro 002 | Google | 1179 ±9 | 8902 |
| 86 | 78–95 | Gemini 2.5 Flash Lite Preview 09 2025 (no-thinking) | Google | 1174 ±10 | 4862 |
| 87 | 80–96 | Gemini 2.0 Flash 001 | Google | 1172 ±7 | 10137 |
| 88 | 85–100 | GPT-4o 2024 05 13 | OpenAI | 1162 ±9 | 23273 |
| 89 | 83–102 | GLM 4.6v | Z.ai | 1161 ±14 | 2455 |
| 90 | 86–100 | Claude 3.5 Sonnet | Anthropic | 1161 ±8 | 10376 |
| 91 | 86–100 | Gemma 3 27b It | Google | 1160 ±8 | 16991 |
| 92 | 86–100 | Mistral Medium 2508 | Mistral | 1159 ±6 | 41647 |
| 93 | 86–104 | Step 1o Turbo 202506 | StepFun | 1156 ±14 | 1899 |
| 94 | 86–103 | Hunyuan Vision 1.5 (thinking) | Tencent | 1156 ±12 | 2430 |
| 95 | 88–102 | Mistral Medium 2505 | Mistral | 1155 ±8 | 10810 |
| 96 | 87–104 | GLM 4.5v | Z.ai | 1153 ±12 | 3249 |
| 97 | 86–108 | Hunyuan Large Vision | Tencent | 1149 ±17 | 1275 |
| 98 | 88–107 | GPT-5 Nano (high) | OpenAI | 1146 ±11 | 3934 |
| 99 | 88–106 | Llama 4 Maverick 17b 128e Instruct | Meta | 1145 ±9 | 6925 |
| 100 | 88–108 | Step 3 | StepFun | 1145 ±12 | 3199 |

## FAQ

### What is the best AI model for vision?

As of 2026-09-28, Anthropic Claude Fable 5 (high) is #1 on the vision board with a score of 1310, rank spread 1–7.

### What is the best open-weights model for vision?

GLM 5.3 Flash (Z.ai), #19 on this board.

### What does this board rank?

Rankings of multimodal models that reason over images: charts, screenshots, OCR and photos.

### How often is it updated?

Whenever new models ship or the rankings change; this board was last updated on 2026-09-28.

Data: [LMArena leaderboard dataset](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset), used under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/); filtered and reasoning tiers merged, scores unchanged.
View the original board: https://arena.ai/leaderboard/vision