READ
READ MORE Predict the properties, effects, and interactions of biological sequences.
THE TECHNOLOGY
Next-Generation Multimodal Genome Language Model
ABOUT
Bringing Biological Data Together At Scale.
0:00 / 0:00
Omnii is trained on raw biological sequences and measurements spanning dozens of modalities, learning how genetic information shapes biological function. This unlocks predictive and generative capabilities across biological applications, from therapeutic design to biosecurity.
TOKENIZATION & ARCHITECTURE
Advancing AI to model the complexity of biology.
Omnii introduces innovations in multimodal representation and model architecture, connecting biological signals across diverse modalities and scales.
01 — A SHARED REPRESENTATION
Biological measurements span heterogeneous data types and formats. Omnii encodes these inputs into a shared representation space, enabling joint learning across modalities.
02 — MULTI-SCALE ARCHITECTURE
Omnii leverages proprietary advances in model architecture design to capture local sequence features and long-range dependencies within a context window up to millions of base pairs.
INSIDE OMNII
How the Model Works
01 / 04
Align
Aligns biological modalities to a common genomic reference, enabling integration across diverse data types.
Represent
Encodes heterogeneous biological inputs into a unified representation through multimodal fusion.
Learn
Learns dependencies within and across modalities, capturing patterns across biological contexts and scales.
APPLY
Generalizes learned representations across biological tasks through targeted post-training for prediction, interpretation, and design.
ADVANTAGE
Generalization Beyond
Specialized Models.
Specialized models — each signal learned on its own, at its own scale.
By learning across modalities, Omnii captures relationships that models specialized for a single data type or task cannot capture in isolation. This broader biological context allows learned relationships to inform predictions across cells, tissues, and conditions.
The result is a general-purpose model for biology, designed to adapt to new tasks by building on knowledge that extends beyond any single application.
Explore the AtlasBROADER CONTEXT
Specialized models — each signal learned on its own, at its own scale.
By learning across modalities, Omnii captures relationships that models specialized for a single data type or task cannot capture in isolation. This broader biological context allows learned relationships to inform predictions across cells, tissues, and conditions.
GENERAL CAPABILITY
The result is a general-purpose model for biology, designed to adapt to new tasks by building on knowledge that extends beyond any single application.
NEXT STEPS
Explore the Research
Read the research behind Omnii’s architecture, capabilities, and results.
Partner with Radical Numerics Inc.
We work with researchers, companies, and institutions to apply Omnii to challenges across disease, therapeutics, and biosecurity. Stay in touch with us below.