---
title: "Vision: model rankings and source evidence · Undominated.ai"
canonical: https://undominated.ai/rankings/arena-vision/
description: "Vision ranked by Arena score. Inspect exact model variants, source dates, measurement limits and downloadable evidence."
---

# Vision: model rankings and source evidence · Undominated.ai

> Vision ranked by Arena score. Inspect exact model variants, source dates, measurement limits and downloadable evidence.

[All rankings](/rankings/) / Vision

# Vision

Human preference ratings for the overall category of this Arena task. Scores from different boards are not interchangeable.

**167 entries** · Arena score · Higher is better

Source publication: 2026-10-19. Retrieved 2026-11-05.

Source: [Arena (lmarena-ai)](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset) · [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/).

Ranks use confidence bounds. Several models can share a rank; the rows within a shared rank are displayed by score.

Vision · Arena score

| Significance rank | Model / evaluated variant | Arena score | 95% confidence interval | Votes |
| --- | --- | --- | --- | --- |
| 1 | Claude Fable 5 (High) anthropic | 1,324.88 | 1,317.85 – 1,331.91 | 14,714 |
| 1 | Claude Opus 5.5 (High) anthropic | 1,320.82 | 1,309.11 – 1,332.54 | 3,035 |
| 1 | Claude Opus 5 (High) anthropic | 1,319.62 | 1,313.01 – 1,326.23 | 17,743 |
| 1 | Claude Fable 5.1 (Max) anthropic | 1,319.49 | 1,310.21 – 1,328.77 | 5,026 |
| 1 | Claude Opus 4.7 anthropic | 1,316.94 | 1,310.39 – 1,323.49 | 23,139 |
| 1 | Claude Opus 4.6 (High) anthropic | 1,315.67 | 1,309.3 – 1,322.05 | 22,286 |
| 1 | Gemini 3.8 Flash (High) google | 1,315.38 | 1,305.64 – 1,325.12 | 4,645 |
| 1 | Gemini 3.7 Flash (High) google | 1,314.23 | 1,305.05 – 1,323.4 | 5,369 |
| 1 | Claude Opus 4.7 (High) anthropic | 1,313.11 | 1,306.58 – 1,319.64 | 22,731 |
| 1 | Qwen3.8-Max alibaba | 1,312.94 | 1,305.49 – 1,320.38 | 9,957 |
| 1 | Claude Opus 4.6 anthropic | 1,312.44 | 1,306.19 – 1,318.68 | 26,841 |
| 2 | Muse Spark 1.3 (Max) meta | 1,308.41 | 1,299.26 – 1,317.57 | 5,392 |
| 2 | Gemini 3.5 Flash (High) google | 1,308.33 | 1,301.51 – 1,315.14 | 13,862 |
| 2 | Gemini 3.5 Flash (Medium) google | 1,306.27 | 1,299.48 – 1,313.06 | 14,061 |
| 2 | Muse Spark meta | 1,305.77 | 1,296.7 – 1,314.85 | 5,865 |
| 3 | Gemini 3 Pro google | 1,304.75 | 1,297.39 – 1,312.12 | 13,623 |
| 1 | Muse Spark 1.2 (xHigh) meta | 1,304.22 | 1,289.89 – 1,318.54 | 1,948 |
| 7 | GPT-5.4 (High) openai | 1,302.63 | 1,296.94 – 1,308.32 | 30,747 |
| 3 | Muse Spark 1.3 (xHigh) meta | 1,301.82 | 1,292.3 – 1,311.35 | 4,897 |
| 7 | Gemini 3.6 Flash (High) google | 1,298.88 | 1,291.01 – 1,306.75 | 8,979 |
| 12 | GPT-5.5 openai | 1,297 | 1,291.24 – 1,302.75 | 28,781 |
| 13 | Gemini 3.1 Pro Preview google | 1,295.78 | 1,290.67 – 1,300.89 | 47,114 |
| 13 | Claude Opus 4.8 (High) anthropic | 1,294.49 | 1,288.04 – 1,300.94 | 18,303 |
| 13 | GLM-5.3-Flash zai | 1,293.87 | 1,287.5 – 1,300.23 | 16,254 |
| 15 | GPT-5.5 (High) openai | 1,293.13 | 1,287.29 – 1,298.97 | 27,067 |
| 14 | Muse Spark 1.1 meta | 1,292.5 | 1,285.61 – 1,299.4 | 12,804 |
| 15 | GPT-5.4 openai | 1,292.03 | 1,285.57 – 1,298.48 | 22,804 |
| 18 | Claude Opus 4.8 anthropic | 1,288.5 | 1,282.02 – 1,294.99 | 18,800 |
| 13 | GPT-6.1 Sol (Max) openai | 1,287.73 | 1,274.87 – 1,300.59 | 2,330 |
| 13 | Claude Sonnet 5.5 (xHigh) anthropic | 1,287.22 | 1,274.71 – 1,299.73 | 2,564 |
| 18 | Grok 4.5 xai | 1,287.12 | 1,280.07 – 1,294.17 | 12,599 |
| 22 | Gemini 3 Flash google | 1,284.95 | 1,279.76 – 1,290.14 | 39,131 |
| 23 | Claude Sonnet 4.6 anthropic | 1,282.76 | 1,276.66 – 1,288.87 | 27,408 |
| 23 | Kimi K2.6 moonshot | 1,282.11 | 1,275.34 – 1,288.88 | 16,368 |
| 24 | GPT-5.6 Sol (xHigh) openai | 1,280.62 | 1,273.33 – 1,287.91 | 11,439 |
| 22 | GPT-6 Astra (Max) openai | 1,279.9 | 1,269.63 – 1,290.17 | 4,012 |
| 28 | Qwen3.7-Plus alibaba | 1,278.15 | 1,271.47 – 1,284.82 | 14,212 |
| 23 | DeepSeek-V4.1-Flash-20260910 (Max) deepseek | 1,277.86 | 1,267.64 – 1,288.08 | 4,100 |
| 28 | Gemma 4 31B google | 1,276.47 | 1,270.8 – 1,282.14 | 40,219 |
| 29 | Dola Seed 2.0 Pro bytedance | 1,274.42 | 1,267.24 – 1,281.6 | 10,972 |
| 31 | Claude Sonnet 5 (High) anthropic | 1,272.55 | 1,265.74 – 1,279.35 | 14,034 |
| 19 | Mistral Large 4 mistral | 1,272.17 | 1,252.95 – 1,291.39 | 1,008 |
| 31 | GPT-5.6 Terra (xHigh) openai | 1,270.97 | 1,263.74 – 1,278.21 | 11,842 |
| 31 | Qwen3.8-27B alibaba | 1,269.46 | 1,261.29 – 1,277.64 | 7,614 |
| 31 | Gemini 3.5 Flash-Lite google | 1,269.07 | 1,261.24 – 1,276.89 | 9,238 |
| 35 | Kimi K2.5 (Thinking) moonshot | 1,269.02 | 1,263.37 – 1,274.67 | 28,249 |
| 35 | GPT-5.2 Chat (20260210) openai | 1,267.83 | 1,261.36 – 1,274.3 | 16,465 |
| 37 | Gemini 3 Flash (Minimal) google | 1,265.71 | 1,260.4 – 1,271.01 | 37,173 |
| 31 | Step 5 Preview stepfun | 1,264.58 | 1,249.54 – 1,279.63 | 1,649 |
| 39 | GLM-5V-Turbo zai | 1,263.91 | 1,258.51 – 1,269.31 | 42,757 |
| 39 | Gemini 2.5 Pro google | 1,263.33 | 1,258.8 – 1,267.86 | 88,755 |
| 37 | Grok 4.6 (High) xai | 1,263.01 | 1,254.7 – 1,271.31 | 7,261 |
| 31 | MiMo-V2.6-Pro xiaomi | 1,262.84 | 1,248.15 – 1,277.53 | 1,754 |
| 39 | Qwen3.5-397B-A17B alibaba | 1,262.47 | 1,257.19 – 1,267.74 | 32,657 |
| 39 | Grok 4.20 Beta (Reasoning, 0309) xai | 1,262.44 | 1,256.39 – 1,268.49 | 26,796 |
| 41 | Gemma 4 26B A4B google | 1,259.55 | 1,253.12 – 1,265.98 | 23,922 |
| 42 | Grok 4.20 Multi-Agent Beta (0309) xai | 1,258.83 | 1,252.74 – 1,264.92 | 24,532 |
| 42 | GPT-5.6 Luna (xHigh) openai | 1,257.15 | 1,249.75 – 1,264.55 | 11,055 |
| 41 | MiMo-V2.6-Flash xiaomi | 1,255.74 | 1,244.92 – 1,266.57 | 3,618 |
| 42 | Kimi K2.5 (Instant) moonshot | 1,253.74 | 1,243.1 – 1,264.37 | 4,111 |
| 44 | Gemini 2.5 Flash Preview (09-2025) google | 1,253.54 | 1,243.83 – 1,263.25 | 4,792 |
| 50 | MiniMax-M3 minimax | 1,251.05 | 1,244.71 – 1,257.4 | 19,844 |
| 50 | GPT-5.5 Instant openai | 1,249.81 | 1,241.22 – 1,258.39 | 7,590 |
| 50 | GPT-5.1 (High) openai | 1,249.51 | 1,241.73 – 1,257.29 | 9,848 |
| 48 | ERNIE-5.0-Preview (1220) baidu | 1,248.45 | 1,237.33 – 1,259.57 | 3,349 |
| 55 | MiMo-V2.5 xiaomi | 1,247.04 | 1,241.23 – 1,252.86 | 29,381 |
| 53 | Qwen3-VL-235B-A22B-Instruct alibaba | 1,246.99 | 1,240.06 – 1,253.92 | 12,724 |
| 56 | GPT-5.2 (High) openai | 1,245.79 | 1,239.75 – 1,251.82 | 19,022 |
| 56 | GPT-5.4 mini (High) openai | 1,244.82 | 1,238.65 – 1,250.99 | 25,734 |
| 56 | Qwen3.5-122B-A10B alibaba | 1,244.5 | 1,237.74 – 1,251.26 | 13,236 |
| 58 | ChatGPT-4o Latest (20250326) openai | 1,243.28 | 1,237.81 – 1,248.75 | 23,400 |
| 59 | Qwen3.5-27B alibaba | 1,241.02 | 1,235.54 – 1,246.49 | 30,272 |
| 59 | Gemini 3.1 Flash-Lite Preview google | 1,240.18 | 1,234.44 – 1,245.92 | 27,657 |
| 56 | GPT-6 Sol (Max) openai | 1,239.95 | 1,228.66 – 1,251.23 | 3,221 |
| 60 | Inkling-Small thinky | 1,236.8 | 1,228.68 – 1,244.91 | 8,074 |
| 67 | Gemini 2.5 Flash google | 1,235.48 | 1,230.95 – 1,240.01 | 60,087 |
| 63 | GPT-5.1 openai | 1,234.9 | 1,227.39 – 1,242.41 | 10,801 |
| 68 | GPT-5 Chat openai | 1,232.19 | 1,225.15 – 1,239.22 | 39,894 |
| 64 | Grok 4.7 (xHigh) xai | 1,230.96 | 1,220.32 – 1,241.59 | 3,736 |
| 73 | Grok 4.3 xai | 1,229.28 | 1,223.46 – 1,235.1 | 27,901 |
| 73 | GPT-5.2 openai | 1,228.61 | 1,222.65 – 1,234.57 | 20,100 |
| 73 | MiMo-V2-Omni xiaomi | 1,227.65 | 1,220.33 – 1,234.96 | 10,650 |
| 74 | Mistral Medium 3.5 mistral | 1,222.45 | 1,212.98 – 1,231.93 | 5,375 |
| 77 | Mistral Large 3 mistral | 1,219.94 | 1,212.19 – 1,227.69 | 9,254 |
| 78 | GPT-6 Luna (Max) openai | 1,216.1 | 1,205.3 – 1,226.91 | 3,662 |
| 81 | o3 (2025-04-16) openai | 1,214.3 | 1,207.77 – 1,220.84 | 45,567 |
| 79 | Qwen-VL-Max (2025-08-13) alibaba | 1,212.57 | 1,200.58 – 1,224.55 | 3,170 |
| 83 | GPT-4.1 (2025-04-14) openai | 1,210.43 | 1,203.76 – 1,217.1 | 40,998 |
| 83 | Grok 4 (0709) xai | 1,210.22 | 1,202.94 – 1,217.5 | 30,173 |
| 83 | GPT-5 (High) openai | 1,208.24 | 1,201.44 – 1,215.03 | 34,445 |
| 83 | Qwen3-VL-235B-A22B-Thinking alibaba | 1,206.96 | 1,194.71 – 1,219.2 | 2,387 |
| 83 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4) nvidia | 1,204.75 | 1,191.88 – 1,217.61 | 2,509 |
| 85 | GPT-5 mini (High) openai | 1,202.23 | 1,194.42 – 1,210.03 | 28,943 |
| 85 | Grok 4.1 Fast (Reasoning) xai | 1,201.17 | 1,194.28 – 1,208.07 | 13,368 |
| 85 | Gemini 2.5 Flash-Lite Preview (Non-Thinking, 09-2025) google | 1,198.29 | 1,188.8 – 1,207.77 | 4,862 |
| 89 | GPT-5.4 nano (High) openai | 1,195.6 | 1,189.31 – 1,201.89 | 25,624 |
| 86 | GPT-4.5 Preview (2025-02-27) openai | 1,194.89 | 1,183.18 – 1,206.61 | 2,925 |
| 90 | o4-mini (2025-04-16) openai | 1,194.01 | 1,186.95 – 1,201.07 | 41,316 |
| 86 | Claude Opus 4 (Thinking 16K, 20250514) anthropic | 1,191.76 | 1,175.79 – 1,207.73 | 1,290 |
| 85 | Claude Sonnet 4 (Thinking 32K, 20250514) anthropic | 1,191.4 | 1,174.46 – 1,208.33 | 1,144 |
| 91 | Gemini 2.5 Flash-Lite Preview (Thinking, 06-17) google | 1,187.47 | 1,180.06 – 1,194.87 | 35,914 |
| 91 | Step-1o Turbo (202506) stepfun | 1,185.94 | 1,171.73 – 1,200.15 | 1,895 |
| 96 | GPT-4.1 mini (2025-04-14) openai | 1,181.41 | 1,173.94 – 1,188.88 | 40,201 |
| 91 | Hunyuan Large Vision tencent | 1,179.89 | 1,163.53 – 1,196.25 | 1,274 |
| 95 | Hunyuan Vision 1.5 Thinking tencent | 1,179.7 | 1,167.57 – 1,191.83 | 2,424 |
| 97 | Step-3 stepfun | 1,176.41 | 1,164.69 – 1,188.14 | 3,193 |
| 97 | Claude Opus 4 (20250514) anthropic | 1,175.62 | 1,162.7 – 1,188.55 | 2,172 |
| 95 | Claude Sonnet 4 (20250514) anthropic | 1,175.42 | 1,161.43 – 1,189.42 | 1,806 |
| 100 | Mistral Medium 3.1 (2508) mistral | 1,172.23 | 1,166.07 – 1,178.38 | 41,611 |
| 98 | Claude 3.7 Sonnet (Thinking 32K, 20250219) anthropic | 1,168.88 | 1,153.35 – 1,184.4 | 1,436 |
| 100 | o1 (2024-12-17) openai | 1,168.37 | 1,157.59 – 1,179.15 | 3,694 |
| 103 | Gemma 3 27B IT google | 1,164.55 | 1,156.53 – 1,172.56 | 16,982 |
| 100 | GLM-4.6V zai | 1,163.94 | 1,150.24 – 1,177.64 | 2,455 |
| 104 | Gemini 1.5 Pro (002) google | 1,161.41 | 1,152.35 – 1,170.48 | 8,902 |
| 104 | GPT-5 nano (High) openai | 1,159.45 | 1,148.41 – 1,170.5 | 3,928 |
| 106 | Gemini 2.0 Flash (001) google | 1,157.87 | 1,150.68 – 1,165.06 | 10,135 |
| 106 | Mistral Medium 3 (2505) mistral | 1,157.01 | 1,148.75 – 1,165.26 | 10,793 |
| 106 | GLM-4.5V zai | 1,154.41 | 1,142.79 – 1,166.04 | 3,243 |
| 105 | Qwen2.5-VL-32B-Instruct alibaba | 1,152.16 | 1,136.78 – 1,167.54 | 1,490 |
| 110 | Claude 3.7 Sonnet (20250219) anthropic | 1,150.92 | 1,141.56 – 1,160.28 | 4,517 |
| 114 | Mistral Small 3.2 (2506) mistral | 1,141.71 | 1,132.65 – 1,150.78 | 10,843 |
| 114 | Llama 4 Maverick 17B 128E Instruct meta | 1,141.71 | 1,132.66 – 1,150.76 | 6,923 |
| 115 | Gemini 1.5 Flash (002) google | 1,140.65 | 1,130.7 – 1,150.6 | 7,241 |
| 118 | GPT-4o (2024-05-13) openai | 1,136.64 | 1,127.85 – 1,145.42 | 23,273 |
| 118 | Mistral Small 3.1 24B Instruct (2503) mistral | 1,136.47 | 1,128.01 – 1,144.93 | 28,490 |
| 121 | Claude 3.5 Sonnet (20241022) anthropic | 1,125.54 | 1,117.64 – 1,133.43 | 10,374 |
| 121 | Step-1o Vision 32K (High Res) stepfun | 1,121.82 | 1,109.16 – 1,134.49 | 2,833 |
| 124 | Claude 3.5 Sonnet (20240620) anthropic | 1,120.33 | 1,110.25 – 1,130.4 | 21,624 |
| 126 | Llama 4 Scout 17B 16E Instruct meta | 1,118.36 | 1,108.92 – 1,127.79 | 6,446 |
| 126 | Qwen2.5-VL-72B-Instruct alibaba | 1,107.39 | 1,096.51 – 1,118.27 | 3,768 |
| 127 | Gemini 2.0 Flash-Lite Preview (02-05) google | 1,100.41 | 1,089.9 – 1,110.93 | 3,991 |
| 130 | Claude 3.5 Haiku (20241022) anthropic | 1,092.53 | 1,076.48 – 1,108.58 | 1,398 |
| 130 | GPT-4 Turbo (2024-04-09) openai | 1,089.88 | 1,078.28 – 1,101.48 | 13,391 |
| 130 | Pixtral Large (2411) mistral | 1,089.4 | 1,080.09 – 1,098.72 | 5,423 |
| 130 | Gemini 1.5 Pro (001) google | 1,088.32 | 1,076.63 – 1,100 | 16,734 |
| 130 | Molmo 2 8B allenai | 1,080.97 | 1,061.37 – 1,100.57 | 1,190 |
| 136 | GPT-4o mini (2024-07-18) openai | 1,065.63 | 1,057.31 – 1,073.95 | 17,341 |
| 134 | GPT-4o (2024-08-06) openai | 1,064.56 | 1,052.08 – 1,077.05 | 3,376 |
| 132 | GPT-4.1 nano (2025-04-14) openai | 1,063.23 | 1,044.94 – 1,081.52 | 1,211 |
| 136 | Qwen-VL-Max (1119) alibaba | 1,057.6 | 1,041.37 – 1,073.84 | 1,422 |
| 138 | Qwen2-VL-72B alibaba | 1,046.95 | 1,036.73 – 1,057.18 | 5,937 |
| 136 | Step-1V 32K stepfun | 1,045.98 | 1,029.28 – 1,062.69 | 1,534 |
| 138 | Gemini 1.5 Flash-8B (001) google | 1,044.11 | 1,033.47 – 1,054.75 | 6,243 |
| 143 | Claude 3 Opus (20240229) anthropic | 1,022.94 | 1,012.58 – 1,033.29 | 15,565 |
| 144 | Molmo 72B (0924) allenai | 1,012.32 | 999.21 – 1,025.42 | 3,048 |
| 144 | Pixtral 12B (2409) mistral | 1,008.58 | 999.02 – 1,018.15 | 7,511 |
| 144 | Gemini 1.5 Flash (001) google | 1,007.97 | 996.1 – 1,019.85 | 13,260 |
| 144 | InternVL2 26B Organization not supplied | 1,004.58 | 991.75 – 1,017.41 | 5,148 |
| 145 | Llama 3.2 90B Vision Instruct meta | 999.67 | 990.41 – 1,008.93 | 8,682 |
| 144 | Hunyuan Standard Vision (2024-12-31) tencent | 999.11 | 977.72 – 1,020.5 | 809 |
| 144 | Aya Vision 32B cohere | 995.6 | 973.59 – 1,017.61 | 847 |
| 145 | Amazon Nova Lite 1.0 amazon | 990.45 | 975.45 – 1,005.45 | 1,854 |
| 145 | Qwen2-VL-7B-Instruct alibaba | 989.93 | 979.37 – 1,000.48 | 5,766 |
| 148 | Claude 3 Sonnet (20240229) anthropic | 983.93 | 972.52 – 995.34 | 12,314 |
| 148 | Amazon Nova Pro 1.0 amazon | 980.46 | 966.58 – 994.34 | 2,335 |
| 148 | Yi Vision Organization not supplied | 977.75 | 959.61 – 995.88 | 1,219 |
| 151 | Llama 3.2 11B Vision Instruct meta | 966.4 | 955.06 – 977.74 | 4,817 |
| 154 | Molmo 7B-D (0924) allenai | 958.72 | 944.91 – 972.53 | 2,815 |
| 156 | Claude 3 Haiku (20240307) anthropic | 950.17 | 937.64 – 962.69 | 13,380 |
| 158 | InternVL2 4B Organization not supplied | 941.33 | 928.8 – 953.87 | 3,703 |
| 156 | nvila-internal-15b-v1 nvidia | 940.17 | 920.29 – 960.05 | 1,077 |
| 158 | LLaVA 1.6 34B Organization not supplied | 936.47 | 924.2 – 948.73 | 4,531 |
| 159 | CogVLM2 Llama3 Chat 19B zai | 927.56 | 912.32 – 942.8 | 1,991 |
| 159 | LLaVA-OneVision Qwen2 72B OV Organization not supplied | 922.84 | 904.84 – 940.84 | 1,321 |
| 161 | MiniCPM-V 2.6 Organization not supplied | 911.64 | 895.65 – 927.62 | 1,987 |
| 166 | Phi-3.5 Vision Instruct microsoft | 851.48 | 835.7 – 867.27 | 2,592 |
| 167 | Phi-3 Vision 128K Instruct microsoft | 811.74 | 793.02 – 830.46 | 1,401 |

## What this ranking can tell you

 - The population is the models in this source board, including variants and models without an available API price.
- Ranks use the published confidence bounds: one plus the number of models whose lower bound exceeds this model’s upper bound. Shared ranks are not proof of equal performance.

## Check the source

The original evidence is retained byte for byte. The downloadable board includes the source URLs, revision and SHA-256 hashes.

[Download this board and provenance →](/data/rankings/arena-vision.json)

 Source files and licences - [metadata · a7a159f557a1](/data/rankings/evidence/a7a159f557a1e594a5d41218fe8d53c63c63daa39b655e51d6fc5b482a17d059.json) · [upstream source ↗](https://huggingface.co/api/datasets/lmarena-ai/leaderboard-dataset)
- [licence · 88a46d472f35](/data/rankings/evidence/88a46d472f35274fb6b7a852c3d16c8ee8007d51b37f290161537bec26895ee4.txt) · [upstream source ↗](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/raw/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/README.md)
- [rows · 56ebfafa4d6b](/data/rankings/evidence/56ebfafa4d6b52a016f797340f9437e3b1c3ac8377d7c4ea2160f356fdc8b96e.parquet) · [upstream source ↗](https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/vision/latest-00000-of-00001.parquet)

[Explore other model types →](/rankings/)[Read the benchmark guide →](/benchmarks/)

## Continue your investigation

 - [Understand the measurements](/benchmarks/)
- [Compare task performance](/best/)
- [Compare embedding models](/embeddings/)
- [Inspect ranking rules](/methodology/)