Model reference · open weights

Virchow2

Embeddings paige-ai Image embed 1 build Non-commercial 110k dl/mo

Virchow2 is an open-weight embedding model from paige-ai. Virchow2 (FP32) weighs 1.3 GB; the smallest configuration that runs it is RTX 3060 12 GB.

Virchow2 is a self-supervised vision transformer developed by paige-ai for image feature extraction in computational pathology.
The model has 631M parameters and was pretrained on 3.1M whole slide histopathology images.
It is released under the CC-BY-NC-ND-4.0 license.

Summary of the paige-ai/Virchow2 model card, 2026-10-01

What it is

What it runs on

Memory and cards for Virchow2 (FP32)

1.3 GBweights, file size

1.1 GBruntime overhead

How it works

Details