CVSBench - FOV Ground-to-Satellite: leaderboard
Metric: Accuracy (%) on FOV-subset ground-to-satellite questions (limited field-of-view street images); four-choice and binary-choice questions over paired street-view and satellite images (Gemini-2.5-flash-generated, human-verified), test split; higher is better. Source: arxiv.org. Saturation forecast: Around November 2027. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 | 56.69 |
| 2 | GPT-5 Chat | 53.5 |
| 3 | InternVL3.5-8B | 52.3 |
| 4 | Gemma 3 4B (IT) | 40.51 |
| 5 | GPT-4o | 36 |
| 6 | Qwen 3 VL 4B Instruct | 26.6 |
| 7 | Qwen 3 VL 8B Instruct | 24 |
Interactive version: theaggregate.ai/benchmark?slug=cvsbench-fov-ground-to-satellite · How It Works · Data refreshed daily, snapshot 2026-09-29.