Today, we're introducing the LGND Research Tier: free Pro-level access to the LGND Embeddings API for qualifying academic researchers. Apply here now. 

Earth embeddings are a promising tool for scientific research. But generating embeddings at scale — or even experimenting with them — requires infrastructure most academic teams can't access. That bottleneck has kept serious geospatial AI work inside a small circle of well-funded institutions. We want to change that.

What You Get

Research Tier members receive full Pro access — 150,000 credits per month, enough to embed Sentinel-2 imagery across the entire Continental US and Europe multiple times over. That also includes:

• Multiple foundation models — including Clay, FarSLIP, AEF, and we’re continually adding more — each encoding spatial and temporal context from imagery
• Embedding Export — pull raw embeddings directly into your modeling stack as GeoParquet via S3, compatible with DuckDB, GeoPandas, or Sedona. Ideal for regression, feature engineering, or model fusion workflows
• MCP Server — query your collections, detect change, and run inference in plain English through any MCP-compatible client like Claude or Cursor
• Code Assistant — keep your agent grounded in current API docs when building on top of LGND; no hallucinated methods, no stale parameters

Check out our developer resources for more information as well.

Our Commitment to Open Science

The Research Tier sits alongside our ongoing open-source work. Earlier this year we published 15.2 billion pre-computed Sentinel-2 embeddings — the full global archive from 2017 through April 2026, embedded with Clay v1.5, free under CC BY 4.0 on Source Cooperative. No paywalls, no proprietary formats. We generated these embeddings as a core part of building LGND, and rather than sit on them, we gave them back. The Research Tier is an extension of that same philosophy.

Apply Now

Are you an academic working with geospatial data? Do you think geographic context from satellite imagery could advance your research — but lack the compute to find out? Apply here.

To learn more about the LGND Embeddings API, visit lgnd.ai/lgnd-docs.