Orb
Orbital MaterialsOrb is Orbital Materials' family of pretrained interatomic potentials for atomic simulations, covering inorganic crystals through the orb-v2 and orb-v3 series and molecular systems through OrbMol. The models predict energies, forces and stresses, expose a per-atom confidence head, integrate with ASE and TorchSim, and can be finetuned on a custom dataset with the packaged script.
Checkpoints are loaded through the orb_models.forcefield.pretrained module rather than declared as a Hub artifact.
Openness
3 high confidence- weights
- open(Apache-2.0 ungated checkpoints loaded through the orb_models.forcefield.pretrained module)
- data
- open(MPTrj, Alexandria, OMat24, OMol25, and OPoly26 training datasets publicly released)
- code
- partial(inference, simulation wrappers, and a finetuning script
- license
- Apache-2.0(OSI)
Orb names a released public dataset behind every checkpoint and licenses the models under Apache-2.0, but the repository ships adaptation code rather than a pretraining run, which keeps it from reaching a fully open, from-scratch release.
- https://raw.githubusercontent.com/orbital-materials/orb-models/main/LICENSE recorded 2026-10-13
Apache License 2.0 body with a 2024 Orbital Materials copyright, read directly rather than from the license endpoint.
- https://raw.githubusercontent.com/orbital-materials/orb-models/main/MODELS.md recorded 2026-10-13
Names the training dataset for each released model: OMol25 and OPoly26 for OrbMol, OMat24 or MPA for v3, MPTraj and Alexandria for v2, each with a public link or reference.
- https://raw.githubusercontent.com/orbital-materials/orb-models/main/README.md recorded 2026-10-13
States that Orb models are licensed under the Apache License, Version 2.0, loads checkpoints from orb_models.forcefield.pretrained with no gate, and documents finetune.py and a finetuning guide but no pretraining entry point.
Adoption
2 high confidenceAdoption is measured from PyPI downloads of orb-models, the declared package that installs the pretrained checkpoints, because PyPI takes precedence over GitHub stars whenever a package is declared.
- https://api.github.com/repos/orbital-materials/orb-models recorded 2026-09-10
stargazers_count: 616; archived: false; fork: false — recorded so the superseded stars signal stays visible.
- https://pypistats.org/api/packages/orb-models/recent recorded 2026-09-10
last_month = 12,043 downloads of the orb-models package.
Capability
2 high confidenceOne attribute. Periodic and non-periodic support is coverage rather than a second scale, the direct and conservative variants are separate checkpoints, and the models predict energies, forces and stresses rather than generating structures.
- https://raw.githubusercontent.com/orbital-materials/orb-models/main/README.md recorded 2026-09-10
Documents finetune.py against any model in ORB_PRETRAINED_MODELS with a dedicated finetuning guide, and describes the models as calculating energies, forces and stresses of atomic systems.
Verified 2026-09-10