AeroGround - View Transformation Reasoning: leaderboard
Metric: Accuracy (%; Spatial Reasoning: select the ground view matching a marked aerial position and heading; aerial-ground collaborative questions from paired simulated UAV and ground views, exact-match scoring of the parsed option or mapping, invalid output scored wrong; one deterministic run). Source: arxiv.org. Saturation forecast: Around June 2027. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 71 |
| 2 | Seed 2.0 Pro | 64 |
| 3 | Qwen 3 VL 32B (Thinking) | 40.3 |
| 4 | Qwen 3 VL 32B Instruct | 38.2 |
| 5 | Qwen 3 VL 8B (Thinking) | 36 |
| 6 | GPT-5.4 | 33.9 |
| 7 | Qwen 3 VL 8B Instruct | 32.8 |
| 8 | Kimi K2.5 | 30.1 |
Interactive version: theaggregate.ai/benchmark?slug=aeroground-view-transformation-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-26.