Every method, with the label the manual gives it.

Analytic nuclear gradients, analytic and finite-difference Hessians, direct and in-core Fock builds.

Over 500 libxc functionals: LDA, GGA, meta-GGA, global and range-separated hybrids, VV10 and double hybrids. Range-separated and VV10 gradients fall back to finite differences and say so in the output.

External PyTorch provider behind the [skala] extra. Energies only, at a fixed geometry.

Native D3(BJ). D4 runs through the dftd4 package; the native D4 backend is parity-validated for H, He and B to Ne only and refuses other elements.

Tuned basis plus dispersion, geometric counterpoise and short-range corrections in one keyword. Calibrated on molecules; the periodic flavours need recalibration.

An opt-in second integral engine for molecules and crystals: exact, RI, multipole (BUPO) or RI-BUPO Coulomb and exact or RI exchange, with an explicit resource budget and the executed provider recorded. The exact build matches libint to 1e-10 Ha; libint stays the default.

RIJCOSX energies within 0.3 mHa of ORCA on glycine/def2-TZVP. The RIJCOSX gradient carries the documented grid-response residual.

libecpint with the Stuttgart-Koeln MDF and LANL2DZ libraries plus per-basis sidecars. ECP-aware analytic gradients on all four molecular drivers.

Conductor-like screening with analytic gradients for HF and DFT. COSMO-RS thermodynamics on the same surface is experimental; solvent with correlated methods is gated.

255 bundled Gaussian sets including the solid-state pob family, on-demand Basis Set Exchange rendering, and the vibe-basis exponent optimiser.

Validated to 1e-9 Ha against PySCF. The dense routes are accuracy references at small-molecule size, not production-size solvers.

Hybrid SCF plus scaled RI-MP2 in one call.

Small-molecule accuracy reference validated to microhartree against PySCF and ORCA. The dense CC iteration remains a production-size memory gate.

Energy-only benchmark implementations pinned against ORCA. CC3 and CCSDT use a dense determinant representation that grows combinatorially.

Local correlation with pair natural orbitals. At the shipped defaults the mean absolute error against canonical CCSD(T) is 0.38 kcal/mol on 17 molecules up to C10H22 (0.19 with TightPNO); state the thresholds when you publish.

Fixed-space variational singles and doubles CI with determinant-space guards.

Multi-root CASCI and state-averaged CASSCF. The state-specific closed-shell CASSCF analytic gradient is finite-difference tight to about 1e-7; PT2 gradients are analytic only in a narrow envelope and finite-difference otherwise.

Exact and near-exact references on a chosen active space with the proper frozen-core dressing; the v2RDM solver has no Q/G semidefinite feedback yet.

A toy matrix-product-state solver cross-checked against full CI; it refuses more than 6 spatial orbitals (12 spin orbitals).

Similarity-transformed Hamiltonian with a research-grade correlator fed to selected CI.

Quasiparticle ionisation potentials and electron affinities, including renormalised GF2 and an open-shell propagator.

RHF, RKS, UHF and UKS references. Excited-state gradients are finite-difference on the RHF/CIS surface only; range-separated kernels are refused.

Independent implementation for H to Xe with COSMO, NEB, molecular dynamics, metadynamics and conical-intersection optimisation. Production within the documented scope.

Screening and preoptimisation with in-house parameters. No DFTB+ parity; absolute energies are not claimable while the repulsive is a placeholder.

Molecular Hamiltonian, analytic gradients and Gamma-periodic derivatives are implemented; external xtb parity and periodic AES electrostatics remain production gates.

Molecular development route. MOPAC heat-of-formation parity is open; periodic PM6 is experimental.

Retirement target: still executable as molecular prescreening, no periodic target, scheduled for removal from the public portfolio.

Recognised spelling that fails closed. The feathered electrostatics are not implemented and PM7 has never been executable in any release.

MSINDO, DFTB, GFN2-xTB and PM6 on a finite cyclic cluster in one, two and three dimensions, with opt-in Madelung embedding.

An external pre-trained model behind the optional [mace] extra, attributed as such. vibe-qc drives its forward pass for energies, forces, optimisation and NEB; it is not a vibe-qc electronic-structure method.

Microhartree parity with PySCF on LiH and MgO, mixed density fitting for dense crystals, closed-shell slab analytic gradients. A 1-D wire is refused and runs as a 3-D cell with transverse vacuum.

Serial route with Gamma RHF/UHF/RKS/UKS, multi-k pure DFT, Gamma analytic gradients and DFT+U, production within that envelope. The run description carries an experimental tag and the grid build warns once while the surrounding family (GAPW defaults, multi-k hybrids) is completed.

Not the automatic default. Molecular-limit Gamma HF and block-construction DFT ship; compact-crystal augmentation validation is pending.

Gamma and multi-k RHF/UHF/RKS/UKS with finite-difference forces and fixed-cell relaxation. Analytic gradients are a preview; the quartet multipole far field is gated.

RI-J from the native GDF loop plus periodic chain-of-spheres exchange on true multi-k meshes.

Vacuum-free two-dimensional Ewald for RHF, RKS and UKS; selected automatically for every slab.

run_periodic_job(method="dlpno-ccsd(t)") on an RHF or UHF GDF reference at Gamma or on a k mesh folded to the Born-von Karman supercell, with the molecular local-correlation engine. Full domains reproduce PySCF KRCCSD(T); accuracy at the shipped thresholds on real crystals is not yet measured.

symmetry_reduce_k=True builds exchange only at the irreducible-wedge representatives and transports it over each star; energies equal the full mesh to roundoff.

Restricted open-shell SCF on the GDF and GPW routes with integer occupations; gradients, smearing and slabs fail closed.

Anderson, Broyden and Kerker mixing with an automatic k-point and smearing recommender.

Dudarev rotationally invariant correction on the GPW route, Gamma and multi-k RKS.

k-path sampling, Gaussian-broadened and Mulliken-projected densities of states, projected band weights.

Energy-resolved crystal orbital overlap and Hamilton populations on the GDF routes, computed with the analysed spin Hamiltonian; skipped with a stated reason on BIPOLE, AICCM and external-XC routes.

k-space Mayer indices; critical-point search and bond-path tracing on periodic and molecular densities.

Finite displacement on the analytic GAPW force at the Gamma point only; multi-k phonons are not wired.

Finite-difference stress on GPW, GAPW and the semiempirical routes. Variable-cell relaxation fails closed on the ab initio routes.

Embedded-surface calculator on a validated 3D Gaussian foundation; thick-slab parity is pending, so absolute energies are qualitative.

Two independent research lines, Gamma-CCM and chi-CCM, from HF through DLPNO correlation on a finite Born-von Karman torus. Every driver warns and no cross-approach result is reportable yet.

Distributed Hartree build behind the [mpi] extra, pending validation on a real multi-rank run.

ASE and native optimisers; molecular IRC by mass-weighted damped steepest descent. Periodic IRC fails closed.

Climbing-image NEB through ASE on any calculator with forces.

Velocity-Verlet, Berendsen and Nose-Hoover dynamics on any surface with forces; well-tempered metadynamics on the same driver; penalty-function MECI optimisation from two tracked states.

Analytic RHF/UHF/RKS/UKS Hessians and finite-difference Hessians for everything else; rigid-rotor harmonic-oscillator thermochemistry.

Coupled-perturbed HF for closed shells.

Emitted into every population sidecar; IAO and IBO analysis with three bonding criteria.

Single-orbital entropies, mutual information and entanglement bond orders on a correlated reduced density matrix; the convenience density wrapper is an approximation.

LMO-EDA and Kitaura-Morokuma terms for two closed-shell Hartree-Fock fragments in the dimer basis, summing to the interaction energy. Kohn-Sham, open-shell and ECP inputs are refused; there is no run_job keyword yet.

NPA raises NotImplementedError until occupancy-weighted natural atomic orbitals land; the NBO search is a provisional Python helper, not a complete NAO/NBO implementation.

Third-order Birch-Murnaghan fits with quality gates; atomisation energies against free-atom references at the same level.

Densities and orbitals for molecules and crystals; Fermi surfaces for XCrySDen.

One QVF archive per run for vibe-view; TREXIO and QCSchema interchange; CIF, POSCAR and extended XYZ readers with symmetrisation; vqfetch for OPTIMADE, Materials Project, COD and NIST data.