Ai2 and Taylor Geospatial join forces on open geospatial AI

Ai2 and Taylor Geospatial are partnering to help conservation, agriculture, and humanitarian organizations build, evaluate, and use geospatial AI. The collaboration will provide open benchmarks, models, and datasets intended to reduce the need for teams to start from scratch.
- Ai2 and Taylor Geospatial announced their partnership at SatSummit in St. Louis.
- Benchmarks of The World will support comparisons between geospatial AI models.
- OlmoEarth offers open Earth-observation models with a no-code platform.
- Fields of The World will be expanded with higher boundary precision, crop type, and yield.
- Datasets, benchmarks, and model weights will be free to use under open licenses.
Open tools for geospatial AI
Announced at SatSummit in St. Louis on November 18, 2026, the partnership focuses on the foundations needed to put geospatial AI to work for the public good. Geospatial AI uses satellite and other Earth observation data to produce maps and insights.
The collaboration will bring together several Ai2 and Taylor Geospatial resources. Benchmarks of The World will help users compare models. Ai2’s OlmoEarth, described by Ai2 as a fully open family of Earth-observation foundation models, includes a no-code platform for fine-tuning and running the models globally.
The partnership will also expand Fields of The World, an open ecosystem for producing a global dataset of farm field boundaries from satellite imagery. The expansion is planned to improve boundary precision and add crop type and yield information.
A response to limited public-sector capacity
Ai2 says geospatial AI is advancing quickly, while efforts to apply it for the public good often lag behind. Few teams have the resources to pretrain, evaluate, and maintain open Earth observation models, according to the organization.
Under the partnership, everything produced, including datasets, benchmarks, and model weights, will be free to use on Source Cooperative and GitHub under open licenses. That approach is intended to make the resulting resources available across domains rather than limiting them to organizations with the capacity to build the underlying systems themselves.
“Open resources change who gets to use geospatial AI,” said Jennifer Marcus, Vice President of Strategic Innovation Programs at Taylor Geospatial. She said the organizations can “build together and deliver open resources faster.”
Practitioners will help shape the benchmarks
The partners say the next stage will focus on the needs of people using geospatial systems in practice. Fields of The World already has a user advisory group of 13 experts from government agencies, UN organizations, nonprofits, universities, and geospatial companies. These members review its data and models and test them in their own work as early pilot partners.
Ai2 and Taylor Geospatial plan to seek similar input for shared benchmarks. Their aim is to test models against the tasks practitioners need to solve in the field, rather than relying only on technical comparisons. The organizations are asking people who build, evaluate, or depend on geospatial foundation models to provide input on which benchmarks should form a shared core.
For users, the partnership could provide a more accessible route from satellite imagery to tested models and usable datasets. For the wider geospatial AI community, the open licensing and practitioner-led evaluation may make it easier to compare systems and build on shared resources.

