LandmarkBias-3K - Region (200 km): leaderboard
LandmarkBias-3K diagnostic geolocation benchmark on 3,000 images selected for misleading or insufficient landmark cues; region-level accuracy accepts predictions within 200 km.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 47.5 | #343 |
| 2 | GPT-4o Mini | 35.77 | #599 |
| 3 | Qwen 2.5 VL 7B | 28.67 | #643 |
| 4 | InternVL2-8B | 11.54 | #817 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=landmarkbias-3k-region-200-km · How It Works · Data refreshed daily, snapshot 2026-10-20.