Bioptimus is building large-scale foundational AI models trained on multimodal biological data to simulate biology from molecules to whole organisms and predict disease outcomes. The company launched H-Optimus-0, an open-source foundation model for pathology trained on millions of images to aid research and diagnosis of diseases such as cancer. Bioptimus plans to broaden its data sources across additional therapeutic areas and to deepen partnerships with pharmaceutical and biotech companies. It will leverage multimodal patient data amassed via ties to Owkin to train its models, while acknowledging the sensitivity and scale of the clinical data required. The company was founded last year by six co-founders, including CEO Jean-Philippe Vert and CTO Rodolphe Jenatton, and is based in France. Financially, Bioptimus has quickly raised capital, bringing total funding to $76M after its latest $41M round following a $35M seed raise. Bioptimus is building large foundational AI models tailored to biological and clinical data with the goal of improving disease diagnosis, precision medicine, and discovering new biomolecules. The company plans to train multimodal models on sensitive clinical datasets amassed through Owkin’s partnerships with academic hospitals. Bioptimus says building these biology models will be capital- and compute-intensive, requiring expensive GPUs and top AI researchers. The team includes Jean-Philippe Vert as co-founder and executive chairman (non-operational), Rodolphe Jenatton as CTO (formerly a senior research scientist at Google), and several co‑founders from Google DeepMind. Bioptimus has signed a partnership with Amazon Web Services to run model training in AWS data centers. The startup recently raised funding to begin model development and data work.

Total raised

$76M

Funding rounds

2

Latest round

Equity

Latest activity

Jan 2025