Radical Numerics is releasing Omnii-1, a genome language model pretrained across 28 biological modalities, including DNA, RNA, proteins, chromatin accessibility, transcription factor binding, histone modifications, DNA methylation, splicing, and gene expression, conditioned on cell type and tissue. It is built to score how a genetic change affects regulation, reconstruct assays that were never measured for a given cell type, and propose DNA for a specified function. With this release, the company reports that bacteriophages designed by Omnii-1 were administered to a patient with an antibiotic-resistant infection under an authorized, physician-led compassionate-use pathway, which it describes as the first clinical use of AI-generated phages. A cocktail containing those designed variants eliminated the patient's antibiotic-resistant isolate in laboratory testing before administration. Radical Numerics is withholding the patient's clinical course and outcome, and it states that a single compassionate-use case does not establish safety or efficacy for broader use. The same pretrained model is also being adapted for regulatory genome prediction, variant nomination in unsolved rare-disease cases, blood-based cancer signal detection, antibody design, cancer vaccine target selection, pathogen screening, and rare-earth binding protein design with Lawrence Livermore National Laboratory. Eric Nguyen, Michael Poli, Stefano Massaroli, and Armin W. Thomas founded the company after building the Evo and Evo 2 genome models at Stanford. Scientists using those earlier models generated the first complete, functional AI-designed bacteriophage genome in the laboratory. Omnii-1 applies that line of work to patient-specific phage design and to screening for biological threats.