Biology is multi-scale
AI models of biology should be, too.
AI is transforming medical research by providing foundation models trained on specific areas: models for genetics, for single-cell, for histology (including H-optimus-1 ). But those areas typically represent a single modality of biology, capturing only one of its scales.
Today, we are taking the next step forward, by creating foundation models that will connect these different modalities and scales.
At Bioptimus, we are building models that—just like biology—are greater than the sum of their parts. Models that understand tissue morphology and gene expression, but more importantly, models that understand and predict how each of these scales will influence the other.
By breaking down data silos and integrating multi-modal biological information with a scalable technology, Bioptimus is unlocking more accurate, better performing AI models—paving the way for faster drug discovery, more precise diagnostics, and the future of biomedical discovery.
The model
M-optimus is being developed progressively, with new modalities and scales integrated in phases.
Each version will expand its capabilities by connecting new data types and making novel linkages between the modalities and scales. Using a bespoke architecture, M-optimus models are being trained end-to-end on both public and proprietary, paired and unpaired datasets.
Our upcoming first release sets the bar high, featuring a powerful baseline of modalities that will unlock unprecedented accuracy and discovery potential.
Early Access
We’re partnering with a limited number of early adopters for first access to the models. This opportunity will allow a few partners to explore its capabilities first-hand, ahead of other market players.
Join us
Visit our Careers page for more details.