AiMolSim
Roadmap

What Is Live, and What Is Not Yet

Live, partial or planned, with the reason beside each. No dates promised, no capability claimed before it passes its checks.

A product and science team at a glass wall with sticky notes arranged as a left-to-right timeline, one phase highlighted in teal
CapabilityStatusDetail
Hartree-Fock (RHF, UHF), STO-3G, 6-31G, def2 to KrLiveEnergy, optimise, frequencies, scan, orbitals
DFT: LDA, BLYP, PBE, HF-LYP, D2 Dispersion, RI-JLiveEnergies, analytic gradients, optimisation, frequencies, scans
Structure Files MOL, SDF, PDBLiveRead and write, V2000 and V3000
Kohn-Sham Wavefunction StabilityLiveEngine 0.7.12: internal, RKS to UKS and real to complex tests
Element Data Re-Verified (CIAAW 2024, AME2020)PartialEngine 0.7.13, rolling out
CheminformaticsLiveProperties, similarity, substructure, clustering, conformers
Model Systems on GridsLiveGrids, models, tunnelling, wavepackets, anharmonic
Thermochemistry (RRHO) and IR SpectraLive298.15 K values, broadened spectra, comparison with experiment
Structures and Structure ChecksLiveMolecules table, parity, close contacts, bonding checks
Clean Up, Atom TypingPartialEngine ready with MMFF94; app steps being switched over
Biomolecules, Crystals, Surfaces, PolymersPlannedStructures sub-tabs, with Materials Science
Molecular Mechanics MMFF94LiveComplete since 0.7.0; UFF awaits verified data
Clusters, Phase 1LivePools of servers with a scheduler and dashboard
Hybrid Functionals (B3LYP, PBE0), RI-J Gradients, XDM DispersionPlannedNext in the build order
Molecular DynamicsPlannedDesign written: thermostats, ensembles
Reaction KineticsPlannedDesign written
Materials, Virtual Screening, QM/MMPlannedNot started or architecture only
MP2, Coupled Cluster, CASSCF, ECPs, TD-DFT, NMR, SolvationPlannedListed as gaps GAP-003 to GAP-009
Semiempirical, Tight-Binding, Machine-Learned PotentialsPlannedListed as gaps GAP-023 and GAP-024
AMBER, CHARMM, OPLS, GAFF, PME, Free EnergyPlannedListed as gaps GAP-015 to GAP-019
Transition-State Search, IRC, NEBPlannedListed as gap GAP-021
Split Computations Across ServersLiveFrequencies, scans and conformers split into parts
Multi-GPU, HPC Over a Fast LANDesign OnlyNeeds suitable hardware to test
Status as of 2026-10-18. Source: the Capability Gap Register and the engine design documents.

Next Engine Releases

The order set by the AiMolSim team on 2026-10-19. Each release ships only after its source papers are read and its results pass their checks, so no dates are given.

0.7.14 B3LYP

Both published variants as explicit choices, never a silent default.

0.7.15 PBE0

Plus Löwdin charges and Mayer bond orders.

0.7.16 Thermochemistry

RRHO thermochemistry computed inside the engine.

0.7.17 IEF-PCM

Implicit solvation by the polarisable continuum model.

0.7.18 TS and IRC

Transition-state search and intrinsic reaction coordinates.

The Capability Gap Register

Inside Chemistry Lab, administrators can open a register of what the engine cannot yet do compared with frontier theory, software, hardware and data. Members see a short Capabilities and Limits view.

  • 51 gaps, each with a severity, a priority and a plan phase; 15 were added on 2026-10-18 for the new domain sub-areas
  • Source references with DOIs checked in Crossref on a recorded date
  • Planned badges appear on module pages, tools and the method ladder
  • Export as CSV, JSON, PDF or Word
Two scientists at a meeting table comparing printed result charts and a laptop showing two nearly identical curves

Roadmap Questions

Why publish a roadmap with gaps?
Because a simulation result is only as trustworthy as what its authors admit. The Capability Gap Register lists 51 gaps against frontier theory, software, hardware and data, each with a severity, a plan phase and dated sources.
Do you give dates?
No. The order is fixed, the dates are not. We would rather ship a verified method late than an unverified one on time.
What comes first?
Hybrid functionals and RI-J gradients, then analytic Hessians with transition-state search and IRC, molecular dynamics with PME, solvation, correlated methods (MP2, coupled cluster, CASSCF), periodic systems, docking and screening, and finally hybrid, coarse-grained and generative methods.
Will DFT results be validated the same way?
Yes. Every new method goes through the same gates: published reference values, cross-engine tests in the separate environment, and a stated tolerance, before it is opened to users.

Run Your Next Molecule Through AiMolSim

Start with a small molecule and a single energy calculation. Every number comes with its method, its basis and where it ran.