French AI startup Bioptimus emerged with foundation-model ambitions in biology, developing spatial gene-expression and tumor-behavior predictors from H&E slides and bulk RNA sequencing. The company says its models are intended to support drug discovery decisions and build toward a so-called “world model” for biology. Bioptimus began in stealth in early 2024 with $35 million in seed funding led by Sofinnova Partners, then followed with an additional $41 million round led by Cathay Innovation. The company states its initial model, H-Optimus-0, is open-source, while H-Optimus-1 is free for academics under license terms for commercial users. A multimodal version (M-Optimus) combines H&E and bulk RNA inputs to predict spatial gene expression across entire slides, and the company claims major pharmaceutical adoption of its earlier tooling. For biotech R&D teams, Bioptimus’ disclosures reinforce the growing use of pathology foundation models to accelerate biomarker interpretation and therapeutic target hypotheses, while also intensifying competition among AI vendors for pharma partnerships.