Multi-Modal Foundation · facebook
OMC25
Predicts properties of organic molecular crystals for chemistry and materials science researchers.
What OMC25 is
Open Molecular Crystals 2025 (OMC25) is a materials modelling resource created by Meta FAIR Chemistry. It provides machine learning models tailored for predicting the physical and chemical properties of organic molecular crystals.
Organic molecular crystals are essential across diverse applications, ranging from pharmaceutical polymorphism to energetic materials and organic light-emitting diodes (OLEDs). The OMC25 baseline models are trained to help researchers study and model these crystalline systems efficiently.
How it works
OMC25 provides baseline model checkpoints, such as eSEN-S, trained on the complete OMC25 dataset. The models integrate into computational pipelines via the fairchem library using a custom Atomic Simulation Environment (ASE) calculator.
Key facts
- Model architecture
- eSEN-S
- Training data
- OMC25 dataset
- Primary application
- Organic molecular crystal property prediction
What you can use it for
- Pharmaceutical Polymorph Analysis
- Predicting the structural and material properties of distinct crystalline forms of pharmaceutical active ingredients.
- Energetic Materials Characterisation
- Evaluating the crystalline behaviour and physical properties of molecular energetic materials.
- OLED Materials Design
- Modelling organic molecular crystal structures relevant to organic light-emitting diodes and optoelectronics.
Run it here
We list this model but cannot run it yet: gated on the Hub — never run past a gate the user did not accept.
Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.
Strengths and limits
Good at
- Trained directly on the full OMC25 dataset specifically focused on organic molecular crystals.
- Interfaces directly with standard computational chemistry workflows through a custom Atomic Simulation Environment (ASE) calculator in fairchem.
Watch out for
- Focuses specifically on organic molecular crystals rather than general inorganic or non-crystalline materials.
- Requires integration with the fairchem repository to run model checkpoints via its ASE calculator interface.
Inputs and outputs
You provide
Sequence or text
A protein or nucleotide sequence, or free text — whichever this model was trained on.
You get back
Mean-pooled representation, one vector per input.
Licence
Check the licence before commercial use · other
This model is gated by its authors — its licence has to be accepted on the source repository before it will run.
Questions
Can I use OMC25 commercially?
Check the licence before commercial use. This model is gated by its authors — its licence has to be accepted on the source repository before it will run.
What model architectures are provided in OMC25?
OMC25 provides baseline checkpoints for the eSEN-S architecture trained on the full OMC25 dataset.
What classes of materials is OMC25 designed to predict?
It is developed for organic molecular crystals, specifically spanning pharmaceutical polymorphs, energetic compounds, and OLED materials.
How is the model executed in simulation software?
Checkpoints are loaded and executed using a custom Atomic Simulation Environment (ASE) calculator available in the fairchem software repository.
How to cite it
Gharakhanyan, V., Barroso-Luque, L., Yang, Y., Shuaibi, M., Michel, K., Levine, D. S., Dzamba, M., Fu, X., Gao, M., Liu, X., et al. (2026). Open Molecular Crystals 2025 (OMC25) dataset and models. Scientific Data. Nature Publishing Group UK London.
Written from OMC25's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.