FourCastNet
No estimate
No hardware requirements for this model
The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.
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
- NVIDIA,NERSC, Lawrence Berkeley National Laboratory,University of Michigan,Rice University,California Institute of Technology,Purdue University
- Organisation type
- Industry,Government,Academia,Academia,Academia,Academia
- Country
- United States of America
- Published
- 22 February 2022
- Authors
- Jaideep Pathak, Shashank Subramanian, Peter Harrington, Sanjeev Raja, Ashesh Chattopadhyay, Morteza Mardani, Thorsten Kurth, David Hall, Zongyi Li, Kamyar Azizzadenesheli, Pedram Hassanzadeh, Karthik Kashinath, Animashree Anandkumar
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Earth science
- Task
- Weather forecasting
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
- tokens
- Epochs
- 130
" The model is pre-trained using a cosine learning-rate schedule with a starting learning rate `1 for 80 epochs. Following the pre-training, the model is fine-tuned for a further 50 epochs" "The training dataset consists of 54020 samples while the validation dataset contains 2920 samples" Global batch size 64 Patch size p × p 8 × 8
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
- 3.5 × 10²⁰ FLOP
- How it was established
- Hardware
312000000000000 FLOP / GPU / sec [A100] * 64 GPUs * 16 hours * 3600 sec / hour * 0.3 [assumed utilization] = 3.4504704e+20 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
- 64
- Wall-clock time
- 16 hours
- Power draw
- 51.5 kW
"The end to end training takes about 16 hours wall-clock time on a cluster of 64 Nvidia A100 GPUs."
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
- Closed — provider access only
- Model access
- Unreleased
- Training code
- Unreleased
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Why it is tracked
- SOTA improvement
- Record confidence
- Confident
"FourCastNet matches the forecasting accuracy of the ECMWF Integrated Forecasting System (IFS), a state-of-the-art Numerical Weather Prediction (NWP) model, at short lead times for large-scale variables, while outperforming IFS for small-scale variables, including precipitation. " "the FourCastNet model has many characteristics that make it superior to the prior SOTA DLWP model"
Sources
Where this record came from and when it was last checked.
- Reference
- FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators
- Last updated
- 28 November 2025
What the numbers mean
What this model is
FourCastNet was published by NVIDIA,NERSC, Lawrence Berkeley National Laboratory,University of Michigan,Rice University,California Institute of Technology,Purdue University, in the country recorded as United States of America, during February 2022. It comes out of an organisation categorised as industry,Government,Academia,Academia,Academia,Academia.
It works in the domain of Earth science, and is recorded as performing the task of weather forecasting.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
What went into building it
Training it took a computation budget of roughly 3.5 × 10²⁰ FLOP, on hardware recorded as NVIDIA A100. That figure measures what producing the model cost, and has no bearing on how fast it answers.
It is tracked in the underlying dataset for one reason in particular: sOTA improvement.
Answers
FourCastNet — common questions
FourCastNet— how many parameters does it have?
No parameter count has been published for it, which is why no memory or speed figure appears on this page.
FourCastNet— who created it?
It was published by NVIDIA,NERSC, Lawrence Berkeley National Laboratory,University of Michigan,Rice University,California Institute of Technology,Purdue University, based in United States of America, an organisation categorised as industry,Government,Academia,Academia,Academia,Academia.
FourCastNet— when was it released?
It was published in February 2022. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
FourCastNet— what is it used for?
It works in the domain of Earth science, and is recorded as handling the task of weather forecasting. These are the areas it was designed around; they describe intent rather than a hard boundary.
FourCastNet— how much compute was used to train it?
Training consumed around 3.5 × 10²⁰ FLOP, on hardware recorded as 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.
FourCastNet— what GPU do I need to run it?
None. This is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.
FourCastNet— is it open source?
No. Its weights have not been published, so it exists only as a service controlled by its owner.
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.