Aerial Visual-Spatial Reasoning on UAV aerial-view benchmark
74.13Overall ScoreGPT-5.5
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GPT-5.5Source Type=Closed-source / API, Setting=vision-language2026.06 | 74.13 | 89 | 54 | 93 | 42 | 92.67 | |
| Gemini-3-Pro-PreviewSource Type=Closed-source / API, Setting=vision-language2026.06 | 71.83 | 90 | 54 | 93 | 33 | 89.17 | |
| Kimi-K2.5Source Type=Closed-source / API, Setting=vision-language2026.06 | 71.8 | 83 | 56.5 | 95 | 35 | 89.5 | |
| Qwen3.6-27BSource Type=Open-source / Local, Setting=vision-language2026.06 | 71.27 | 77 | 55 | 94 | 44 | 86.33 | |
| Qwen3.6-35B-A3BSource Type=Open-source / Local, Setting=vision-language2026.06 | 69 | 78 | 53 | 87 | 43 | 84 | |
| Grok-4.20-0309-ReasoningSource Type=Closed-source / API, Setting=vision-language2026.06 | 68.57 | 79 | 54.5 | 91 | 33 | 85.33 | |
| Moonshot-VL-128K-Vision-PreviewSource Type=Closed-source / API, Setting=vision-language2026.06 | 67.53 | 82 | 53 | 84 | 32 | 86.67 | |
| Claude-Opus-4-7Source Type=Closed-source / API, Setting=vision-language2026.06 | 67.23 | 82 | 56 | 83 | 30 | 85.17 | |
| Claude-Sonnet-4-6Source Type=Closed-source / API, Setting=vision-language2026.06 | 67.07 | 79 | 55 | 85 | 31 | 85.33 | |
| GPT-5.2Source Type=Closed-source / API, Setting=vision-language2026.06 | 65.87 | 74 | 53 | 85 | 32 | 85.33 | |
| GPT-4.1Source Type=Closed-source / API, Setting=vision-language2026.06 | 63.13 | 68 | 63 | 79 | 24 | 81.67 | |
| Qwen3-VL-PlusSource Type=Closed-source / API, Setting=vision-language2026.06 | 56.87 | 61 | 44 | 78 | 23 | 78.33 | |
| Claude-Haiku-4-5-20251001Source Type=Closed-source / API, Setting=vision-language2026.06 | 51.33 | 55 | 43.5 | 67 | 24 | 67.17 | |
| Qwen3-VL-Flash-2025-10-15Source Type=Closed-source / API, Setting=vision-language2026.06 | 49.87 | 50 | 37.5 | 67 | 23 | 71.83 |