
BioEmu-1
Deep-learning emulator of protein structural ensembles, orders of magnitude faster than MD.
About
BioEmu-1 is a biomolecular emulator from Microsoft Research AI for Science that predicts equilibrium ensembles of protein structures at a speed orders of magnitude faster than traditional molecular dynamics simulations. Trained on over 200 milliseconds of MD simulation data plus static structure databases and experimental protein stabilities, it samples functionally relevant conformational changes — cryptic pocket formation, local unfolding, large-scale domain rearrangements — with relative free-energy errors around 1 kcal/mol against millisecond-timescale MD and lab measurements. It is released under the MIT license with a companion BioEmu Benchmark.
The honest picture: BioEmu-1 emulates ensembles rather than running true physics, so its outputs are only as good as its training distribution — novel folds far from the data remain risky. It is a research tool, not a drug: it generates hypotheses for experimental testing, and the 'thousands of structures per hour on one GPU' headline assumes batched, well-behaved inputs.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 57; Trending score 82/100 from researcher assessment.