GeoSpark

Verifiable spatial reasoning for large language models.

Distances, topology, coordinate reference systems and heights computed by a geometry engine and official geodetic models, not estimated by a language model. Open source, self-hostable, and usable by any model that can call a tool.

GeoSpark provides language models with tools that compute spatial answers rather than estimate them: geodesic distances, exact topology, official datum and height conversions, geocoding, elevation, routing, open map data and live environmental data. Each result carries its sources, licences, units and caveats, and a result that cannot be computed reliably is reported as such rather than approximated. GeoSpark also checks users' own datasets (vector files, LiDAR point clouds and elevation models) for the silent coordinate and height errors that most often corrupt spatial analyses.

Contents: Motivation and evaluation history · Capabilities · Access and installation · Usage · Command-line interface · Evaluation · Validation against authoritative sources · Limitations · Self-hosting · Release history · Data attribution · Development · Citation

Motivation and evaluation history

GeoSpark is evaluated against the models in current use, and its scope follows the gaps that those evaluations reveal.

• GeoSpark Bench v2. Across nine models of the time, four of the five small open-weight models (3.8B–9B parameters) got at most one numeric distance question in twenty right, and frontier APIs ranged from 25% to 95% correct. With GeoSpark's tools, every model that issued valid tool calls answered 76–96% correctly; a 7B model running on a CPU improved from 5% to 76%.
• Re-evaluation with current models (October 2026). On a 38-question sample of the same benchmark, this time with the coordinates written into each question, Claude Opus 5.5 and GPT-6.1 answer without tools and miss one question between them; with GeoSpark's tools a 4B open model on a workstation matches them. Once the coordinates are supplied, distances are no longer the limiting problem.
• Identification of the next gap. Both frontier models get datum shifts, plate motion and map projections right to the millimetre, yet fail on conversions that require a national grid file: ellipsoidal (GNSS) heights to AHD, ODN, NZVD2016 or NAVD88, and British National Grid through OSTN15. Answering directly, including within the providers' own code sandboxes (which have no internet access), GPT-6.1 declined and Opus 5.5 estimated, with errors of up to 9.9 m.

... View full README on GitHub