# H-optimus (Bioptimus)

Source: https://onco.cc/technologies/h-optimus/  
OnCo record `h-optimus` (Technology). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

An open 1.1-billion-parameter pathology model from a French startup, among the strongest on public benchmarks.

## Summary

H-optimus is a ViT-giant vision transformer for histopathology from the French company Bioptimus, pretrained with the DINOv2 self-supervised method so it learns tissue features without labels. H-optimus-0 (2024) was trained on hundreds of millions of tiles from 500k slides and led public benchmarks on release; H-optimus-1 followed. The weights are open and available for research and commercial licensing, which makes the model a common backbone for groups building their own biomarker or diagnostic classifiers. The gap is clinical: it has less clinical validation than commercial products that have gone through regulatory review, so users must validate downstream tasks themselves. For a newcomer: it is a strong, freely downloadable pathology model that other tools can be built on, not a diagnostic product in itself.

## Fields

- Kind: Technology
- Status: emerging
- Last checked: 2026-09-08
- Tags: foundation-model; pathology
- Principle: H-optimus is a ViT-giant pretrained with DINOv2.
- Since: 2024
- Strengths: Open weights; Benchmark leader on release
- Limitations: Less clinical validation than commercial products

## Sources

- Hugging Face model card: https://huggingface.co/bioptimus/H-optimus-0

## Connected records

- fronts: [AI & Computation](https://onco.cc/fronts/ai-computation/)
- technologies: [Pathology & radiology foundation models](https://onco.cc/technologies/pathology-foundation-model/)
- companies: [Bioptimus](https://onco.cc/companies/bioptimus/)
- roadmaps: [AI in oncology roadmap: pattern readers → foundation models → agents in the workflow](https://onco.cc/roadmaps/ai-oncology-roadmap/)

---
JSON: https://onco.cc/api/v1/entities/h-optimus.json