More details on training and benchmarking Sage 2.3.0 are available in our preprint: bsky.app/profile/open...

And in Python:
>>> from openff.toolkit import Molecule, ForceField
>>> forcefield = ForceField("openff-2.3.0.offxml")
>>> molecule = Molecule.from_smiles("CCO")
>>> interchange = forcefield.create_interchange(molecule.to_topology())

The easiest way to try out Sage 2.3.0 is by creating a new environment with the latest OpenFF ForceFields:

micromamba create -n openff -c conda-forge "openff-toolkit>=0.17" "openff-forcefields==2026.01.0"

micromamba activate openff

Sage 2.3.0 maintains or improves performance on all of our benchmarks except solvation free energies in nonaqueous solvents. Please try it out and let us know if you find any substantial improvements, regressions, or other issues in comparison to Sage 2.2.1!

AshGC is used in our latest force field, Sage 2.3.0

Thomas Steinbrecher joins as an elected representative of the industry partners who fund our efforts and provide advice and direction. He has been an active and influential voice in this community, and we appreciate his commitment to the success of the project.

Danny Cole joins as an additional PI, in recognition of the central role his independent research has been playing in advancing the science and infrastructure of Open Force Field.

Our seven-member Governing Board includes two elected representatives from supporting Partners and five Principal Investigators. The Governing Board makes strategic operational decisions and oversees expenditures.

The Open Force Field Consortium is a pre-competitive, industry-funded effort to build more accurate force fields and to improve predictive power of computational drug discovery techniques.

Installation instructions and usage examples are available in the repository README at github.com/openforcefie...

Pablo v0.2.0 includes expanded custom residue definition capabilities, including the ability to define "anonymous" residues that don't rely on atom names. The workflow demonstrates parameterizing a post-translationally modified protein system and running a short OpenMM simulation.

The updated workflow uses a new tool (openff-pablo v0.2.0) for loading modified proteins from PDB files and the new prototype Rosemary force field (OpenFF 3.0.0 alpha 0) for a self-consistent treatment of both canonical and noncanonical protein residues.

The scientific work is still ongoing, but we’re releasing a prototype version so you can try it out now and learn how it will fit into your workflow once it’s fully ready for production.

Years of research on modeling proteins are beginning to pay off in a new force field that can do something no other publicly available force field can do: accurately model proteins, peptides, and general organic molecules with one consistent set of parameters.

Compared to Sage 2.2.1, Sage 2.3.0-rc2 splits torsions so that a single torsional parameter only covers a central bond with a single multiplicity. It also adds some bond and angle types to improve performance of targeted chemistries, such as small rings.