Sources, operations, assumptions and limitations remain visible.
Company · Synta LGM
Spatial intelligence,
built by operators and engine builders.
LGM is jointly owned and developed by SynPlanet and GUIDE Digital. The partnership connects planetary spatial AI with real city operations, simulation and enterprise delivery.
880 vCPU and 12,960 GB RAM in the secured 2026 capacity.
API · MCP · GIS · workflows · digital twins.
Our thesis
A map locates an object.
LGM helps a system understand what it means.
Geospatial work remains fragmented across large datasets, specialist tools, infrastructure and manual analysis. LGM turns those parts into one reusable intelligence layer for agents, applications, engineers and decision-makers.
- 01Data
- 02Intelligence
- 03Decision
- 04Simulation
- 05Action
Built in partnership
City-scale execution.
Planetary ambition.
The model is only the core. Operational context, simulation and multiple distribution surfaces make the technology useful in the real world.
01Development partner
SynPlanet
Planetary spatial AI & simulation
Builds the planetary data model, synthetic environments and the path from geospatial intelligence into interactive worlds.
- SynTa spatial intelligence
- SynTerra for Unreal and web
- Large-scale synthetic environments
02Development partner
GUIDE Digital
City-scale operations & delivery
Brings real municipal workflows, digital-twin operations and enterprise delivery across complex spatial systems.
- GIS, BIM, IoT and video workflows
- Unreal and browser delivery
- Sovereign and on-premises deployments
Proof in use
Research that ends in a spatial decision—not a generic answer.
In the Tbilisi study, LGM orchestrated at least 117 MCP calls, processed 109,316 rows and raster samples, compared ten street corridors and built a shortlist of district opportunities.
- 109K
- records and samples
- 10
- street corridors
- 240
- planning cells
Why this team can deliver
Three capabilities that rarely live in one company.
The partnership can develop the model, connect it to real operational data and deliver the result into the environments where decisions are made.
Operate complex territories
GUIDE reports a city-scale digital twin in operation since 2022, connecting GIS, BIM, IoT, video and analytical layers.
Build interactive worlds
SynPlanet carries geospatial data into browser and Unreal environments through SynTerra and synthetic-world systems.
Ship reusable engines
LGM development applies an engine-first approach: one intelligence core distributed through API, MCP, GIS and digital twins.
Leadership & development
Spatial AI, operations, simulation and platform engineering.
A cross-domain team is essential because LGM must work as data infrastructure, an AI system and a usable product at the same time.
Dr Pavel Entin
Co-founder & CEO · SynPlanet
Platform vision and spatial-AI ecosystem development.
Dr Andrey Kirsanov
Co-founder & CTO · SynPlanet · CSO GUIDE Digital
AI, computer graphics and the technical bridge between both partners.
Tatiana Stack
Co-founder & COO · SynPlanet
Operations, compliance and enterprise product delivery.
Aleksandr Golubev
Head of LGM Development
Serial platform and simulation-engine builder leading the LGM product architecture.
Steve Bannigan
Chairman · SynPlanet
Company building, financial discipline and capital-markets experience.
Romain Janil
Chief Product Officer · SynPlanet
Large-area simulation, procedural cities and real-time environments.
Infrastructure capacity
Backed for the compute-intensive stage.
LGM is an AWS Partner with a secured 2026 compute and storage allocation for data processing, model inference and enterprise pilots. Further GPU-intensive stages are planned for 2027 and 2028.
Future stages are shown as cumulative planned capacity and remain under discussion.- 12,960 GB
- System RAM
- 104 GB
- GPU VRAM · 2 GPUs
- 50 TB
- S3 storage
3× x8i.64xlarge · c7gn.12xlarge · r8g.8xlarge · g6.4xlarge · p5.4xlarge
- 15,008 GB
- System RAM
- 744 GB
- GPU VRAM · 10 GPUs
- 150 TB
- S3 storage
Adds p5.48xlarge and 100 TB S3
- 21,152 GB
- System RAM
- 4,320 GB
- GPU VRAM · 26 GPUs
- 150 TB
- S3 storage
Adds p6-b200.48xlarge and p6-b300.48xlarge
Work with us
Bring LGM to your territory, platform or data product.
Start with the partnership path closest to your work. We can begin with a usage-based connection or develop a dedicated enterprise system.
Governments & territories
Build a sovereign geospatial intelligence layer for planning, monitoring and municipal operations.
Platforms & AI products
Add raster, vector, GeoRAG and spatial computation through API or MCP.
Data & technology partners
Make valuable datasets, models or simulation engines available inside LGM workflows.
Large Geospatial Model