TerraMind
No estimate
No hardware requirements for this model
This model's weights are open, but no parameter count has been published for it. Every memory and speed figure starts from that number, so we would rather show nothing than a fabricated estimate.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- IBM,Forschungszentrum Julich,European Space Agency (ESA),NASA
- Organisation type
- Industry,Government,Government,Government
- Country
- United States of America, Germany, Multinational
- Published
- 15 April 2025
- Authors
- Johannes Jakubik, Felix Yang, Benedikt Blumenstiel, Erik Scheurer, Rocco Sedona, Stefano Maurogiovanni, Jente Bosmans, Nikolaos Dionelis, Valerio Marsocci, Niklas Kopp, Rahul Ramachandran, Paolo Fraccaro, Thomas Brunschwiler, Gabriele Cavallaro, Juan Bernabe-Moreno, Nicolas Longépé
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Earth science, Vision
- Task
- Image captioning, Flood Mapping, Crop Mapping / Segmentation, Wildfire Mapping, Cloud monitoring / analysis
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Training data
- 500,000,000,000 tokens
"The model was pre-trained on 500B tokens from 9M spatiotemporally aligned multimodal samples from the TerraMesh dataset."
Training compute
The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.
- Training compute
- 3.1 × 10²¹ FLOP
- How it was established
- Hardware
312000000000000 FLOP/GPU/sec * 9216 GPU-hours * 3600 sec / hour * 0.3 [assumed utilization] = 3.10542336e+21 FLOP
The training run
What it physically took to train: which chips, how many, for how long, and what that drew from the wall.
- Training hardware
- NVIDIA A100
- Chips used
- 32
- Chip-hours
- 9,216
- Wall-clock time
- 288 hours (12 days)
- Power draw
- 25.1 kW
"Overall, the training of TerraMindv1-B took 12 days on 32 A100 GPUs, i.e., 9’216 GPU hours"
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Open — downloadable
- Model access
- Open weights (unrestricted)
- Training code
- Unreleased
- Hugging Face
- ibm-esa-geospatial
Models and code have been open-sourced at https://huggingface.co/ibm-esa-geospatial and https://github.com/ibm/terramind ("This repo presents code examples for fine-tuning TerraMind"). Apache 2.0 https://huggingface.co/ibm-esa-geospatial/TerraMind-1.0-base
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
Sources
Where this record came from and when it was last checked.
- Reference
- TerraMind: Large-Scale Generative Multimodality for Earth Observation
- Last updated
- 28 November 2025
What the numbers mean
About this model
TerraMind was published by IBM,Forschungszentrum Julich,European Space Agency (ESA),NASA, in United States of America, in April 2025. industry,Government,Government,Government is the category the publisher falls under.
It works in Earth science, Vision, and is recorded as doing image captioning, Flood Mapping, Crop Mapping / Segmentation, Wildfire Mapping, Cloud monitoring / analysis.
Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded. It is published under the ibm-esa-geospatial organisation on Hugging Face.
Training and provenance
The training run consumed about 3.1 × 10²¹ FLOP, on NVIDIA A100. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
The training set ran to roughly 500,000,000,000 tokens.
Answers
TerraMind — common questions
What GPU do I need to run TerraMind?
We cannot say. TerraMind has open weights, but no parameter count has been published for it, and every memory and speed calculation starts from that number. We would rather show nothing than a fabricated estimate.
Is TerraMind open source?
Its weights are published, so TerraMind can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.
How many parameters does TerraMind have?
No parameter count has been published for TerraMind, which is why no memory or speed figure appears on this page.
Who created TerraMind?
TerraMind was published by IBM,Forschungszentrum Julich,European Space Agency (ESA),NASA, based in United States of America, categorised as industry,Government,Government,Government.
When was TerraMind released?
TerraMind was published in April 2025.
What is TerraMind used for?
TerraMind works in Earth science, Vision, and is recorded as handling image captioning, Flood Mapping, Crop Mapping / Segmentation, Wildfire Mapping, Cloud monitoring / analysis. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
Where can I download TerraMind?
Its weights are published under the ibm-esa-geospatial organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.
How much compute was used to train TerraMind?
Around 3.1 × 10²¹ FLOP, on NVIDIA A100. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.