EarthSpatialBench (Direct) - Angle Estimation: leaderboard
Metric: Mean absolute error in degrees of the estimated azimuth between two referenced objects on EarthSpatialBench exact-angle questions on 512x512 SatlasPretrain Earth images at about 1 m resolution, direct prompting; lower is better. Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 30B A3B Instruct | 117.2 | #365 |
| 2 | Qwen 3 VL 30B A3B (Thinking) | 117.98 | #338 (Qwen 3 VL 30B A3B) |
| 3 | GPT-5 | 119.26 | #91 |
| 4 | Claude 3.5 Sonnet | 130.18 | #337 |
| 5 | Gemini 2.5 Pro | 134.47 | #145 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=earthspatialbench-direct-angle-estimation · How It Works · Data refreshed daily, snapshot 2026-10-11.