API Reference

MoC-JEPA is a pretrained framework — Moment-driven Charge Density Joint Embedding Predictive Architecture — that maps a structure into per-node latents, then into a charge-density grid you can drop into PySCF as an SCF initial guess. Billing is metered on Processed Atoms for every density prediction. The API is operated by 本征实验室 (Braketlab Inc.). All requests go over HTTPS to https://api.moc-jepa.ai/v1.

1. 01

   Upload structure

   Crystal .cif / .xyz, or molecule .pdb / .xyz.

2. 02

   Encode latents

   Backbone returns per-node latents as .h5.

3. 03

   Predict density

   DEN head writes .cube (default) or .chgcar.

4. 04

   Accelerate SCF

   Feed the grid to PySCF as init_guess.

Authentication

Authenticate with a secret key in the Authorization header. Generate keys from your dashboard. Never expose a secret key in client-side code.

Pretrained Backbones

Two independently pretrained backbones share the same architecture and versioning rule. Pick the one that matches your chemistry.

Density Prediction Models

Density is the shipping application. Each DEN head is bound to one backbone major line and is the model you pay API tokens against. Future heads (Band Gap Prediction, Potential Energy Prediction) will reuse the same slot.

Naming & Versioning

Checkpoints follow a single pattern. The three-letter task slot is reserved so new applications can land without renaming the backbone.

• major — latent space is not backward compatible; every downstream head must be retrained.
• minor — same architecture, more or cleaner data; latents stay compatible.
• A head at DEN-v1.x binds to backbone major line v1.x.
• API ids are the lowercase kebab form. Unpinned ids (moc-jepa-m1-den) resolve to the latest minor within the current major. Pin with moc-jepa-m1-den-v1.0.
• Reserved task codes: GAP Band Gap Prediction PES Potential Energy Prediction — documented, not shipping.

Encode Latents

Maps an input structure into the MoC-JEPA latent space and returns per-node latents as an .h5 file. The model is equivariant under the Euclidean group E(3), so for any rotation R the latents transform as z(Rx) = R · z(x).

Each Processed Atom yields a 256-dimensional latent vector. The full tensor has shape (N, 256) and dtype float32.

Request body

Predict Density

Runs the recommended Density Prediction Model on a latents file and returns a charge-density grid. This is the billable call: one Processed Atom is counted per atom in the encoded structure.

Request body

One-shot Density

Convenience endpoint that encodes latents and predicts density in one request. Equivalent to calling /v1/latents then /v1/density. Prefer the two-step path when you want to reuse latents across multiple heads.

Request body

Accelerating SCF with PySCF

Download the density grid, then initialize the SCF density matrix from that cube instead of a minao or atomic guess. The helper projects the real-space density onto the chosen basis and returns a density matrix you pass to mf.kernel.

1. Call /v1/density or /v1/density:direct and save the .cube.
2. Build the PySCF molecule with the same geometry you uploaded.
3. Construct dm0 via initial_guess_from_cube.
4. Run mf.kernel(dm0=dm0).

File Formats

Rate Limits

Limits are enforced per Processed Atom, not per request. Both /v1/latents and /v1/density count against the same quota. The default tier allows 10,000,000 Processed Atoms per month and 200 requests per minute. Contact us for research-scale throughput.