{"architecture":{"backbone":"ViT-H/14","family":"Vision Transformer","input":"224\u00d7224 tile at 20\u00d7","output":"2,560-dim embedding (concatenated class token and mean patch token)","params":"632M","pretraining":"DINOv2 self-supervised"},"article":null,"cancer_slugs":["pan-cancer"],"category":"pathology","confidence":"high","datasets":[],"developer":"Paige / Memorial Sloan Kettering Cancer Center","evaluation":[{"benchmark":"Pan-cancer detection \u2014 17 cancer types (9 common, 8 rare), specimen level, trained on Virchow tile embeddings","external":false,"metric":"AUROC","source":"https://www.nature.com/articles/s41591-024-03141-0","value":"0.95 (Vorontsov E et al., Nat Med 2024; PMID 39039250)"}],"hf":{"downloads":7616,"fetched_at":"2026-09-09T21:33:08Z","gated":"auto","last_modified":"2024-09-03","library":"timm","license":"apache-2.0","likes":76,"pipeline_tag":"image-feature-extraction"},"kind":"foundation","license":"cc-by-nc-nd-4.0","limitations":"- Single-institution pretraining; scanner (Leica Aperio) dominance.\n- Non-commercial licence.\n- Superseded by Virchow2 for most uses.","links":{"demo":null,"docs":null,"doi":"10.1038/s41591-024-03141-0","github":null,"huggingface":"https://huggingface.co/paige-ai/Virchow","paper":"https://www.nature.com/articles/s41591-024-03141-0","pmid":null},"modalities":["histopathology"],"name":"Virchow","notable_uses":"Pan-cancer detection AUC 0.95 overall in the paper, with strong performance on rare cancers where task-specific models lack data.","openness":"gated-weights","regulatory":{"intended_use_en":"Research release of the encoder; Paige's cleared products are separate.","intended_use_pl":"Badawcze wydanie enkodera; produkty Paige z dopuszczeniem to osobne wyroby.","source_url":"https://huggingface.co/paige-ai/Virchow","status":"research-only"},"regulatory_status":"research-only","release_date":"2024-07-22","run_snippet":"# generic timm loader \u2014 check the model card for the exact init args and image normalisation\nimport timm, torch\nfrom huggingface_hub import login\nlogin()  # gated repos: accept the licence on huggingface.co first\nmodel = timm.create_model('hf-hub:paige-ai/Virchow', pretrained=True)\nmodel.eval()\ncfg = timm.data.resolve_data_config({}, model=model)\ntransform = timm.data.create_transform(**cfg)\n# emb = model(transform(tile).unsqueeze(0))  # 1 x D tile embedding\n","settings":["diagnosis","basic-research"],"slug":"virchow","sources":[{"label":"Vorontsov E et al. A foundation model for clinical-grade computational pathology and rare cancers detection. Nat Med 2024","url":"https://www.nature.com/articles/s41591-024-03141-0"},{"label":"Hugging Face \u2014 paige-ai/Virchow","url":"https://huggingface.co/paige-ai/Virchow"}],"summary":"632-million-parameter vision transformer pretrained on 1.5 million whole-slide images from about 100,000 patients \u2014 the largest pathology pretraining set at its release \u2014 and used to build a pan-cancer detection model covering 17 cancer types, including rare ones.","tasks":["feature-extraction","classification","detection"],"training":{"institutions":"Memorial Sloan Kettering Cancer Center (New York, USA)","populations":"single tertiary cancer centre","size":"1.5M WSIs","summary_en":"1.5 million H&E whole-slide images from about 100,000 patients at Memorial Sloan Kettering Cancer Center, spanning 17 tissue types.","summary_pl":"1,5 mln preparat\u00f3w H&E od oko\u0142o 100 000 pacjent\u00f3w Memorial Sloan Kettering Cancer Center, 17 typ\u00f3w tkanek."},"updated_at":"2026-09-09T21:33:09.703713","url":"/ai-oncology/models/virchow","usage":{"library":"timm","notes_en":"Gated on Hugging Face under a non-commercial licence; Virchow2 (mixed magnification, 3.1M slides) is the successor and usually the better default."},"verified_at":"2026-09-05T22:26:01.133062","verified_by":"editorial","version":"Virchow (v1)","what_it_does":"Tile-level embeddings (class token plus mean of patch tokens) for slide-level aggregation; Paige used them for a clinical-grade pan-cancer detector and for biomarker prediction."}