Paige, a leader in clinical AI for cancer, has launched the second generation of its Virchow foundation models, setting a new benchmark in AI-driven pathology.

Developed in collaboration with Microsoft, Virchow2 and Virchow2G are now the largest and most advanced AI models in clinical pathology. Trained on over 3 million pathology slides from 800 labs across 45 countries, the models are designed to enhance cancer diagnosis and treatment by providing a more comprehensive understanding of cancer at a cellular level.

The Virchow2G model, with 1.8 billion parameters, is three times larger than its predecessors, making it the most extensive pathology model to date. The models were trained using deidentified data from over 225,000 patients, ensuring a diverse representation across gender, race, and ethnicity, which aids in creating more precise and personalized cancer diagnostics.

Paige’s collaboration with Microsoft has been critical in leveraging advanced supercomputing resources, allowing for a scale of AI training that surpasses previous industry standards. "Virchow’s immense scale unlocks key information that can drive groundbreaking innovations," said Thomas Fuchs, founder and chief scientist at Paige, emphasizing the potential for more accurate diagnostics and targeted treatments.

Paige has also developed AI modules for pharmaceutical and life sciences companies, enabling more effective therapeutic targeting and clinical trial design. Virchow2 is also available on Paige’s OpenPFM Suite on Hugging Face for non-commercial research, expanding its accessibility to the scientific community.