Garden-AI Launches Almanac of Matter Models: Invoke 25+ MLIPs in Two Lines

BenBlaiszik · x · 2026-08-14

Garden-AI announced the launch of the "Almanac of Matter Models," aiming to make using the latest matter models as easy as calling an LLM. The project argues that solving cutting-edge scientific problems requires ensembles of Machine Learning Interatomic Potentials (MLIPs) rather than relying on a single model.

Currently, the platform has pre-installed and continuously verified over 25 MLIP families (including MACE, CHGNet, MatterSim, etc.). These models are deployed at DOE leadership computing facilities like Argonne National Laboratory. Researchers can now swap models or form a committee of them with just a two-line code change, eliminating complex setup costs.

Related event: Garden-AI Launches Almanac of Matter Models(2 posts)→

Original post →

More from Research

Research channel →