Coordinate conversion and projection benchmark for geospatial AI
How well AI models do coordinate conversion and projection on satellite, aerial and map data: 1 set with results, scored the same way for every model.
Sets
| Set | Index | Tasks of this type | Best score on the set | Models |
|---|---|---|---|---|
| GPSBench closed book | Spatial Intelligence Index | all | 93 | 9 |
Top models
Best score on GPSBench closed book, the set most models have run, 0 to 100.
| Rank | Model | Developer | Score |
|---|---|---|---|
| 1 | Step 5 Preview | StepFun | 93 |
| 2 | Nemotron 3.5 Super 120B | NVIDIA | 90 |
| 3 | Nemotron 3.5 Super VL 120B | NVIDIA | 89 |
| 4 | Mistral Large 4 | Mistral AI | 83 |
| 5 | Nemotron 3.5 Lightning 30B | NVIDIA | 78 |
| 6 | Qwen 3.6 35B-A3B | Alibaba (Qwen) | 74 |
| 7 | Muse Glimmer 30B | Meta | 73 |
| 8 | Nemotron 3 Nano Omni 30B | NVIDIA | 64 |
| 9 | Laguna XS 2.1 | Poolside | 43 |
Every task type · Every evaluation · Methodology
Frequently asked questions
Which AI model is best at coordinate conversion and projection on geospatial data?
Step 5 Preview leads, on the sets below. Best score on GPSBench closed book, the set most models have run, 0 to 100.
What is this coordinate conversion and projection benchmark?
Every task in our sets that asks for coordinate conversion and projection, in GIS and Earth observation work, with each set's own score. It is a view over the sets, not an index of its own.
Which index does it count in?
Each set counts in one index: the Spatial Intelligence Index, as the table shows.
