AI Frontier

Builds AI foundation models, based on AlphaFold-derived structural biology, to predict how drug molecules and antibodies bind and behave, aiming to design new therapeutics computationally instead of relying on traditional wet-lab screening.

Builds an end-to-end AI-designed therapeutics platform, combining genome-wide perturbation data with generative models to design new drugs computationally, positioned as a direct peer to Isomorphic Labs in scale and ambition.

Builds foundation models (now Chai-3) for de novo design of antibodies and other biologics, aiming to replace parts of traditional wet-lab screening with generative structure-function prediction, similar in spirit to Isomorphic Labs but focused on biologics rather than small molecules.

An AI drug discovery company combining generative chemistry, biology, and clinical-trial-design models (Pharma.AI) to design drug candidates and advance them into human trials.

Deep learning model predicting the 3D structure of proteins and, since AlphaFold3, their interactions with DNA, RNA, ligands, and other molecules, using a diffusion-based architecture.

AlphaGenome is a sequence-to-function model predicting gene expression, chromatin accessibility, histone modification, transcription factor binding, and 3D genome folding from DNA sequences up to ~1 million base pairs. The Atlas applies it at genome scale, precomputing predicted effects for roughly 9 billion possible single-nucleotide variants across the human genome, packaged as a browsable ~1-petabyte dataset.

A DNA/genomic foundation model (40 billion parameters, up to 1-million-base-pair context) trained on roughly 9 trillion base pairs spanning all domains of life, able to both interpret and generate genetic sequences.

An open platform and toolkit for building, training, and deploying biomolecular AI models (proteins, RNA, genomics), hosting or interoperating with models like Evo 2 and ESM, positioned as infrastructure for the whole field rather than a therapeutics company of its own.

OpenAI's first frontier reasoning model purpose-built for biology, drug discovery, and translational medicine, fine-tuned for genomics, proteomics, and molecular biology tasks, named after Rosalind Franklin.