AI models grounded in the physical sciences and built for real-world impact.
AQCat is a machine learning interatomic potential, trained on 13.5 million quantum chemistry calculations across 47,000 catalyst systems, that predicts catalyst performance up to 20,000 times faster than conventional physics-based methods while approaching their accuracy.
It uniquely models spin polarization for magnetic, earth-abundant metals such as iron, nickel, and cobalt, filling a data gap that leading catalyst datasets and models leave open while matching their accuracy elsewhere.
AQPotency is a Large Quantitative Model for potency prediction and virtual screening that ranks protein-ligand pairs by predicted binding strength (pIC50) without requiring a 3D crystal structure of the target.
It runs on standard CPUs, scores pairs in seconds for about a dollar per thousand, and works best as a fast ranking method. Every prediction carries an uncertainty estimate and an applicability-domain score, so scientists can tell which results fall inside the chemical space the model has learned.
AQCat Dispersion expands SandboxAQ’s catalyst modeling capabilities to large, commercially vital molecules, including C5+ fuels and plastics precursors.
It is designed to help researchers computationally screen catalysts across a wider range of reactions before committing to more costly laboratory experimentation
AQADME is an ADME and pharmacokinetic property-prediction model for candidate molecules.
It is designed to help research teams prioritize compounds for hit identification, hit-to-lead, and lead optimization.
AQCat is a machine learning interatomic potential, trained on 13.5 million quantum chemistry calculations across 47,000 catalyst systems, that predicts catalyst performance up to 20,000 times faster than conventional physics-based methods while approaching their accuracy.
It uniquely models spin polarization for magnetic, earth-abundant metals such as iron, nickel, and cobalt, filling a data gap that leading catalyst datasets and models leave open while matching their accuracy elsewhere.
AQCat Dispersion expands SandboxAQ’s catalyst modeling capabilities to large, commercially vital molecules, including C5+ fuels and plastics precursors.
It is designed to help researchers computationally screen catalysts across a wider range of reactions before committing to more costly laboratory experimentation
AQPotency is a Large Quantitative Model for potency prediction and virtual screening that ranks protein-ligand pairs by predicted binding strength (pIC50) without requiring a 3D crystal structure of the target.
It runs on standard CPUs, scores pairs in seconds for about a dollar per thousand, and works best as a fast ranking method. Every prediction carries an uncertainty estimate and an applicability-domain score, so scientists can tell which results fall inside the chemical space the model has learned.
AQADME is an ADME and pharmacokinetic property-prediction model for candidate molecules.
It is designed to help research teams prioritize compounds for hit identification, hit-to-lead, and lead optimization.
*Offer available to first-time users of the SandboxAQ MCP server only. New users receive a one-time $2,000 promotional credit applied at sign-up, redeemable solely against usage of the SandboxAQ MCP server and valid across any available model. The credit must be used within 30 days of activating your trial; any unused portion expires at the end of that period and does not roll over. The credit has no cash value, is non-transferable, non-refundable, and cannot be exchanged for cash or applied to any other product, subscription, or fee. A valid payment method is required to activate your trial. Once your $2,000 credit is exhausted, or once the 30-day period ends — whichever comes first — standard usage-based pricing applies and your payment method will be charged for continued use of the SandboxAQ MCP server. Limit one credit per individual. SandboxAQ may modify, suspend, or end this offer at any time. Offer subject to the SandboxAQ Terms & Conditions.
