Large Geospatial Model

LGM, at a glance.

LGM (Large Geospatial Model) is a backend geospatial intelligence platform that combines raster and vector data, spatial computation, GeoRAG and AI orchestration. It helps people and applications investigate territories and return evidence, indicators and map-ready results.

Product identity

One spatial backend.

LGM is the product name; Synta LGM is the identity used on synta.global. The platform is jointly developed by SynPlanet and GUIDE Digital. LGM is not simply one language model: specialized models are one part of its development roadmap.

About LGM

From question to evidence

Define an area and a question, find relevant accessible data, run spatial operations and inspect the result. GeoRAG supplies geographic evidence; the orchestrator coordinates models and tools.

Use it to investigate service gaps, compare candidate locations or screen environmental and infrastructure conditions. Results support investigation; they do not replace local validation or professional decisions.

Choose an interface

Different ways into LGM.

01Browser

Web GIS & Chat

Investigate a territory on a map with LGM chat. The application at gis.synta.global brings together Mapbox maps, Cesium 3D and Google Earth content.

Explore LGM GIS
02Account

Platform Dashboard

The account surface for registration, sign-in, API keys, uploaded layers and billing. Available controls depend on the rollout and account configuration.

Explore the Platform
03AI agents

MCP Server

Expose LGM data and spatial tools to compatible agents through an authenticated remote MCP connection. Use the endpoint and credentials issued for your account.

Connect through MCP
04Applications

API

Integrate geospatial data and computation into your own application. API keys identify access and usage; supported operations depend on the service.

Explore the API
05Private beta

QGIS Plugin

Bring LGM layers and analysis into desktop GIS. The current public site describes private-beta access, not an unrestricted marketplace release.

Review QGIS access
06Roadmap

Model Provider

Future specialized and fine-tuned geospatial models for partner catalogs. This is a development direction, not a claim of currently published model endpoints.

Explore the model roadmap

Availability

Available work. Clear boundaries.

Core services · rolling out

Raster, vector, GeoRAG and orchestration services are being introduced progressively. GeoAI deployments already use project uploads, shared collections and public Overture/Australia layers through LGM.

Explore tools and GeoRAG

Account controls · staged

Platform brings account access, keys, layers and billing together. Do not assume that every organization, payment, budget or permission feature is enabled for every account. Confirm the scope of your deployment.

Review Platform capabilities

Models & vision · roadmap

Specialized fine-tuned models and Computer Vision object detection from imagery are development directions. Additional GIS, engineering and digital-twin connectors are not all released integrations.

Review integration status

Data and commercial scope

Know the inputs and the limits.

Public, shared and private context

The data foundation includes OpenStreetMap/Overture vector context, elevation, surface classification, soil and STAC catalog discovery. Collection coverage, resolution and freshness vary by source.

COG is used for raster data; GeoParquet for vectors; GeoJSON carries application-ready geometry; STAC describes discoverable assets. Catalog access does not imply unrestricted access to every underlying dataset.

Explore data and formats

Access, usage and licensing

A project uses its uploaded layers, collections shared with it and available public layers. Dataset licenses, attribution and redistribution conditions remain source-specific. Open data is not the same as free LGM processing.

The commercial model is usage-based, with custom enterprise engagements. Public rates and included limits should be checked on the pricing page and in the account; this overview does not promise a free usage allowance.

Review pricing information

Continue with a concrete task

Read the evidence. Choose a path.