NVIDIA

Brain MRI synthesis in 3D across T1w, T2w, FLAIR and SWI, whole-brain or skull-stripped, at up to 512x512x256 voxels in 30 sampling steps.

Flow-matching generative model for de novo atomistic protein binder design against protein and small-molecule targets, including carbohydrate binders.

Longitudinal chest X-ray vision-language model that reads a follow-up radiograph against its prior study and forecasts how findings will change.

Genomic foundation model training framework whose joint-embedding predictive objective learns functional representations of masked DNA, not tokens.

Partially latent flow-matching model for de novo protein design, jointly generating sequence and all-atom structure for proteins up to 800 residues.

Generative transformer foundation model for continuous glucose monitoring, forecasting glycemia and stratifying health risk from raw glucose traces.

Open-source Apache-2.0 reproduction of AlphaFold3 that predicts all-atom structures of proteins, RNA, DNA, small molecules, and their complexes.

Synthetic 3D MRI generation spanning brain, prostate, breast and abdomen, with the T1, T2 or FLAIR contrast selected at sampling time.

Molecular graph transformer for ADMET property prediction, pretrained on 11M molecules under a joint SMILES-reconstruction and contrastive objective.

All-atom structure prediction for arbitrary biomolecular complexes of proteins, nucleic acids, and ligands, with code and weights under a BSD license.

RNA-seq splice junction detection pairing transcriptome-guided alignment with a fine-tuned DNABERT scorer, reaching a mean F1 of 0.971.

Codon-resolution language models trained on 130 million coding sequences from 20,000 species, learning codon rules for translation and mRNA stability.

Functional ultrasound reconstruction with neural operators, rebuilding power Doppler brain images from 10-15% of the usual compounded frames.

Flow-matching protein backbone generator with hierarchical fold-class conditioning, reaching 99.0% designability and chains up to 800 residues.

3D CT image synthesis with paired anatomical masks, driven by a rectified-flow latent diffusion model that samples a whole volume in 30 steps.

Fragment-based molecular generation by masked discrete diffusion over SAFE strings, covering de novo design, linker design, and lead optimization.

Antibody-antigen binding affinity prediction and sequence optimization, pre-trained on 7.5 million quantitative yeast-display affinity measurements.

All-atom protein dynamics simulator that bridges consecutive MD snapshots directly using SO(3)-equivariant two-sided stochastic interpolants.

3D latent diffusion model for synthetic CT, generating volumes up to 512x512x768 voxels with paired masks across 127 anatomical structures.

Chest X-ray conversational assistant that fine-tunes LLaVA-Med on instruction data enriched with predictions from expert radiograph classifiers.

3D vision-transformer foundation model for multimodal neuroimage segmentation, pretrained self-supervised on brain MRI from 41,400 participants.

Medical image segmentation foundation model for 3D CT and MRI, covering 127 anatomical classes automatically plus interactive point-prompt refinement.

Generalist cell instance segmentation for microscopy, pairing a SAM ViT-B encoder with flow-field decoding trained on 15K annotated images.

Brain MRI foundation model pretrained with masked image modeling on roughly 57,000 multi-contrast head scans for brain tumor diagnosis.

Abdominal CT segmentation model driven by CLIP text embeddings, covering 25 organs and 6 tumor types with zero-shot extension to new categories.

Small-molecule generation over SMILES through a fixed-size latent space dense enough that gradient-free search over it optimizes chemical properties.

Chemical language model for small-molecule drug discovery, embedding SMILES for property prediction and sampling new molecules from its latent space.