About
Bioptimus is a Paris-founded AI company on a mission to build the world model of biology — a universal AI foundation model that integrates multi-modal, multi-scale biological information to break down existing data silos in science and medicine. By training on over 2 billion images and data from 4,000+ clinical practices, Bioptimus develops powerful foundation models that capture the underlying laws of biology at a level of complexity previously unattainable. Their flagship offering, H-Optimus-1, is the leading foundation model for histology and ranks #1 out of 22 foundation models for digital pathology. It has been used in landmark case studies including breast cancer recurrence risk prediction and spatial transcriptomics workflows for IBD research. M-Optimus represents their broader world model of biology initiative. The platform is built on four critical pillars: high-quality patient data sourced through clinical partnerships, a world-class team of ML and biology researchers, best-in-class GPU infrastructure, and enterprise-grade data security. With over 1 million downloads across their model portfolio and approximately 100 scientific papers published annually using their models, Bioptimus has established significant traction in the global research community. They have raised $76M in funding and partner with leading institutions including MIT. Bioptimus is aimed at pharmaceutical companies, academic research institutions, clinical labs, and biotech organizations seeking to leverage AI for biological discovery and medical innovation.
Key Features
- H-Optimus-1 Histology Model: The #1 ranked foundation model for digital pathology out of 22 competing models, enabling high-accuracy histological analysis and clinical prediction tasks.
- M-Optimus World Biology Model: A broad-scope world model of biology designed to integrate multi-modal, multi-scale biological data into a unified, queryable knowledge base.
- STELA Spatial Biology Atlas: The world's largest clinically linked spatial biology atlas, built in partnership with 10x Genomics and Broad Clinical Labs, enabling spatial transcriptomics research at unprecedented scale.
- Secure, Scalable AI Platform: Enterprise-grade infrastructure with best-in-class GPU compute, robust data security, and access to over 2 billion training images from 4,000+ clinical practices.
- Research-Ready Model Downloads: Foundation models available for download with nearly 1 million cumulative downloads, supporting ~100 scientific publications per year across the global research community.
Use Cases
- Predicting breast cancer recurrence risk from histopathology slides using H-Optimus-1 integrated with the STAMP framework
- Automating pathology report generation for clinical and research institutions using foundation model-based AI pipelines
- Accelerating spatial transcriptomics workflows in inflammatory bowel disease (IBD) and other complex disease research
- Enabling cancer progression-free survival prediction in partnership with leading academic medical institutions like MIT
- Training and fine-tuning custom biological AI models on proprietary datasets using Bioptimus's secure, high-performance infrastructure
Pros
- Top-Ranked Pathology Model: H-Optimus-1 is the most consistently highly ranked model across 22 digital pathology foundation models, providing researchers with state-of-the-art performance.
- Strong Institutional Adoption: Partnerships with MIT, 10x Genomics, Broad Clinical Labs, and 4,000+ clinical practices demonstrate wide trust and real-world validation.
- Comprehensive Biological Data Coverage: Trained on 2 billion images across multi-modal, multi-scale biological data, enabling generalization across diverse research domains.
- Active Research Community: With ~100 scientific papers published annually using their models, there is a thriving ecosystem of research built on the Bioptimus platform.
Cons
- Highly Specialized Domain: Bioptimus models are designed specifically for biology and biomedicine, limiting applicability outside the life sciences.
- Limited Public Pricing Transparency: Pricing and access terms for enterprise or clinical use are not publicly detailed on the website, requiring direct engagement with their team.
- Early-Stage Platform: Founded in 2024, the platform and model ecosystem are still maturing, with some features and model offerings still under active development.
Frequently Asked Questions
What is H-Optimus-1?
H-Optimus-1 is Bioptimus's flagship foundation model for histology and digital pathology. It is ranked #1 out of 22 foundation models in the field and is used for tasks such as cancer recurrence prediction, pathology report generation, and spatial transcriptomics analysis.
What is STELA?
STELA (Spatial Transcriptomics and Expression Linked Atlas) is the world's largest clinically linked spatial biology atlas, announced in March 2026 in partnership with 10x Genomics and Broad Clinical Labs. It provides researchers with an unprecedented resource for spatial biology research.
Who is Bioptimus designed for?
Bioptimus is designed for pharmaceutical companies, academic research institutions, clinical laboratories, biotech organizations, and computational biologists who want to leverage AI to accelerate biological and biomedical discoveries.
How can I access Bioptimus models?
Bioptimus models are available for download via their platform and have accumulated nearly 1 million downloads. Researchers can access models through the Bioptimus platform, and enterprise or clinical integrations can be arranged through direct partnership.
How is patient data security handled?
Bioptimus employs enhanced data security measures that allow scientists and researchers to safely access, analyze, and manage vast amounts of sensitive biological and clinical data within a secure, scalable computing environment.
