All Competitors

Energy-based flow matching for 3D molecular structure, using an idempotent predict-and-refine map for protein backbone generation and ligand docking.

Patient-level representation learning from scRNA-seq: a transformer set encoder with a diffusion decoder, fine-tuned for clinical prediction.

Codon language model trained with synonymous-codon-constrained masking, so its embeddings encode nucleotide-level signal, not amino acid identity.

Predicts residue-residue dynamic contact maps from a single sequence, matching molecular dynamics ensemble methods at a fraction of the compute.

Generalist MRI vision-language foundation model that handles reconstruction, segmentation, abnormality detection, and report generation in one model.

Low-dose CT denoising across dose levels and body regions from one checkpoint, conditioning residual diffusion on learned dose and anatomy embeddings.

Simplex diffusion model for discrete sequence generation, with released checkpoints for DNA enhancer design and de novo protein sequence design.

Slide-level pathology foundation model fusing five frozen tile encoders into one whole-slide embedding for biomarker and therapy response prediction.

Biological reasoning language model that post-trains Qwen3 with reinforcement learning from verifiable rewards for drug-discovery tasks.

Vision-language foundation model for kidney cancer CT, covering zero-shot malignancy diagnosis, report generation, and recurrence risk prediction.

Generative model that restores cytoplasm-enriched genes lost in snRNA-seq, recovering cell-cell communication signals from raw nuclear counts.

Protein-ligand binding model that embeds ligands, pockets, and sequences in hyperbolic space, unifying virtual screening and affinity ranking.

Single-cell foundation model for yeast that injects regulatory network priors into transformer attention for zero-shot and fine-tuned analysis.

Multi-organ segmentation network jointly trained on 10 CT and OCT datasets, reaching 0.83 mean Dice on the RETOUCH retinal fluid challenge.

Reasoning language model post-trained on virtual cell simulations, answering questions about gene perturbations and their effects in natural language.

All-atom protein structure diffusion models for motif scaffolding and hotspot-conditioned complex generation, at 22M parameters.

Protein function prediction via compressed in-context learning on a sequence-structure language model, cutting 751-token demonstrations to under 16.

Dermatology foundation model pretrained on 432,776 skin images, covering malignancy classification, severity grading, and lesion segmentation.

Multimodal foundation model predicting genome-wide binding of chromatin-associated proteins from protein sequence, DNA sequence, and chromatin state.

B-cell epitope predictor pairing CNN and Transformer branches over protein language model embeddings to score linear and conformational epitopes.

Ice-binding protein classifiers over frozen ESM-2 embeddings, separating antifreeze from ice-nucleation proteins across bacterial proteomes.

Self-supervised bioacoustic audio encoder that turns animal-sound recordings into transferable embeddings for species classification and detection.

Genomic DNA foundation model using ELECTRA-style replaced-token detection and single-nucleotide tokenization; 93M parameters rival 2.5B baselines.