Train a model on your data

Upload a CSV or SDF. Pick a base model. We handle the GPUs, evaluation, checkpointing, and model-card compliance. Your weights, your IP.

1
Base model
2
Dataset
3
Config
4
Review
5
Train

Choose a base foundation model

MoLFormer-XL

ADMET, quantum, physical-chemistry. 112M params. Best for HTS data.

Uni-Mol

3D conformer-aware. Best for binding affinity, pose ranking, crystallography.

ChemCrow + Llama-3

Agent reasoning with tool-use. Best for de-novo design and multi-step workflows.