Material Science
Materials science underpins progress in energy, electronics, and mobility by engineering properties at the atomic and molecular scale.
Breakthroughs in semiconductors, batteries, and nanomaterials drive technological transformation and sustainability.
Paramus INOR
The Paramus.ai Materials Science Solution, driven by Thorsten Gressling, integrates AI-based simulation, structure–property prediction, and accelerated materials discovery pipelines.
$899 USD*
* named user, per year, terms apply.
Included
The materials science package from Paramus contains:
Models
Universal Potentials & AI Models
Foundation model for inorganic materials (89 elements, MPtrj)
Multi-head model: energy + forces + stress + MLFF fine-tuning
Trained on OMAT24 for bulk, surface, and defect simulations
Microsoft deep-learning atomistic model for materials
Orbital Materials universal potential for periodic systems
Neural potential for organic/elemental-organic molecules
Data
Datasets (Materials & Structures)
Open-access crystal structures for organic/inorganic compounds
134k organic molecules: HOMO, LUMO, gaps, dipole, Cv
QM9 + IR, Raman, UV-Vis spectra and tensorial properties
Amorphous silicon structures from GAP-driven MD
76 molecules with CCSD(T)/CBS atomization energies
Battery
Battery Simulation & Test Data
Physics-based battery models (SPMe, DFN) with experiment workflows
Battery cell test-data analysis (cycles, dQ/dV, efficiencies)
Builders
Structure Builders
Build crystals, supercells, and slabs; export CIF/POSCAR
Simulation & HPC
Quantum Chemistry & Molecular Dynamics
Massively parallel MD for materials, alloys, ceramics
MD engine optimized for periodic systems with GPU
Mixed Gaussian/plane-wave DFT for periodic systems and surfaces
Quantum chemistry: HF, DFT, MP2, CCSD(T), SAPT
Scalable QC for HF, DFT, plane-wave DFT, AIMD (MPI)
Ab initio QC: HF, DFT, MP2, MCSCF, CI, relativistic
Scalable MD for large biomolecular and materials systems
Build initial MD configurations: boxes, slabs, interfaces
Atomic Simulation Environment: Python interface to DFT/MD codes
Semi-empirical QC for fast screening of large systems
6 foundation/ML potentials (MACE-MP-0, MACE-MPA-0, MACE-OMAT-0, MatterSim, ORB, AIMNet2), 10 simulation engines (LAMMPS, GROMACS, CP2K, PSI4, NWChem, GAMESS, NAMD, Packmol, ASE, xTB), 5 crystal/thermo datasets (COD, QM9, QM9S, a-Si-24, MSR-ACC/TAE25), 2 battery apps (PyBaMM, cellpy) and Crystal Builder – all local under PARAMUS OS.
Yes. PyBaMM runs physics-based battery models (SPMe, DFN) and cellpy analyzes your cycler data (cycles, dQ/dV, efficiencies) – both run locally and are callable from your AI assistant.
No. All models and engines run locally on CPU under PARAMUS OS. No CUDA, no cloud account – your structures stay on your machine.
Crystal Builder creates crystals, supercells and slabs and exports CIF/POSCAR; ASE and Packmol prepare MD boxes and interfaces; CP2K, LAMMPS and the ML potentials run the calculation.
Through the Model Context Protocol (MCP). Claude Desktop, VS Code or Cursor call the tools directly: structure building, ML potentials, battery simulation, dataset lookups.