CVSBench - Viewpoint Arrow: leaderboard
Metric: Accuracy (%) on View-Arrow: choose the camera-pose arrow on a satellite image that matches a given street-view image; 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 2034. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 8B Instruct | 42.3 |
| 2 | Qwen 3 VL 4B Instruct | 40.8 |
| 3 | InternVL3.5-8B | 40.1 |
| 4 | GPT-5 Chat | 40.1 |
| 5 | Gemma 3 4B (IT) | 30.13 |
| 6 | GPT-4o | 30 |
| 7 | DeepSeek V3.2 | 26.13 |
Interactive version: theaggregate.ai/benchmark?slug=cvsbench-viewpoint-arrow · How It Works · Data refreshed daily, snapshot 2026-09-29.