OLMo Earth

2026-07-25 · Source: A Geodyssey – Geoscience Text Analytics and Enterprise Search Research · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Earth Observation · Depth: Intermediate, quick

Summary

OLMoEarth, an open-source Earth-observation AI system developed by Ai2, has released its V1.1, serving as the earth image counterpart to Ai2's OLMo language models. It features a family of foundation models, trained on approximately 10 terabytes of global satellite and sensor data, ranging from 1.4 million to 300 million parameters, all utilizing a vision-transformer architecture. These models are designed to convert raw Earth-observation signals into actionable analysis, such as land cover, crop type, and deforestation. The system also includes an online platform with a Studio for model building, a Viewer for map exploration, a workflow engine, and an API, aiming to enable governments, NGOs, and local communities to use the technology without deep AI expertise. Its distinguishing feature is its complete openness, providing model weights, training code, datasets, and evaluation stacks, with a license prohibiting use for extraction industries.

Key takeaway

For governments, NGOs, and local communities seeking to integrate Earth observation data without extensive AI expertise, OLMoEarth offers a compelling open-source solution. Its transparent models, training data, and code allow for full inspection, customization, and self-hosting, mitigating vendor lock-in. You should explore its no-code Studio and API to integrate advanced geospatial analysis into your environmental monitoring and planning initiatives.

Key insights

OLMoEarth democratizes Earth observation AI through open-source models and a user-friendly platform.

Principles

• Openness enables inspection, rebuilding, and self-hosting.
• Small, efficient models ensure broad accessibility.
• No-code interfaces lower AI expertise barriers.

Method

The platform offers a Studio for model building/fine-tuning, a Viewer for map exploration, and a workflow engine for analysis.

In practice

• Updating global mangrove maps.
• Detecting Amazon deforestation.
• Predicting wildfire risk.

Topics

• Earth Observation
• Foundation Models
• Open-Source AI
• Satellite Imagery
• Geospatial AI
• Environmental Monitoring

Best for: Computer Vision Engineer, AI Scientist, Research Scientist, Domain Expert, Machine Learning Engineer, AI Engineer

Related on AIssential

• Introducing OlmoEarth embeddings: Custom embedding exports from OlmoEarth Studio for downstream analysis — OlmoEarth embeddings offer a cost-effective, label-efficient way to analyze Earth observation data for diverse downstream tasks.
• OlmoEarth v1.1: A more efficient family of Earth observation models — OlmoEarth v1.1 achieves 3x compute cost reduction by optimizing transformer token sequence length without performance loss.
• Strait of Hormuz crisis drives demand for commercial geospatial intelligence — Geospatial intelligence platforms combining diverse open-source data are crucial for real-time maritime domain awareness in contested zones.
• How our open-source AI model SpeciesNet is helping to promote wildlife conservation — SpeciesNet is an open-source AI model accelerating wildlife monitoring and conservation through automated image identification.
• Ph.D. student opening in Sweden on Earth Observation, Data Science, and AI for poverty estimation — The AI & Global Development Lab uses AI and Earth Observation to analyze human development and poverty dynamics from 1984 onwards.
• NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather — NVIDIA's Earth-2 provides an open, accelerated AI software stack for comprehensive weather and climate prediction.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by A Geodyssey – Geoscience Text Analytics and Enterprise Search Research.