{"entity":{"id":"atlas-aignostics","kind":"technology","name":"Atlas (Aignostics, Mayo Clinic, Charité)","aka":[],"tldr":"Atlas is a pathology foundation model trained on 1.2 million slides from two of the world's largest hospitals.","summary":"Atlas is a pathology foundation model from Aignostics developed with Mayo Clinic and Charité, a ViT-H vision transformer trained with RudolfV-style self-supervision. Its distinguishing feature is the training set: 1.2 million slides from two of the world's largest hospitals, spanning diverse scanners and stains, which is intended to make the learned features robust to the variation seen in routine practice. The 2025 arXiv paper reported top scores on public pathology benchmarks. The model is proprietary, so outside groups cannot inspect or fine-tune the weights and must rely on the company's evaluations, and clinical validation of downstream tasks remains to be published. For a newcomer: Atlas is a hospital-trained slide model that scores well on benchmarks but is not openly available.","status":"emerging","asOf":"2026-09-08","links":[{"label":"Atlas (arXiv 2025)","url":"https://arxiv.org/abs/2501.05409"}],"tags":["foundation-model","pathology"],"related":[],"cancers":[],"sections":["ai-computation"],"technologies":["pathology-foundation-model"],"targets":[],"drugs":[],"companies":["aignostics"],"institutions":["mayo-clinic","charite"],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":[],"keyPapers":[],"journals":[],"dependsOn":[],"notes":[],"principle":"Atlas is a ViT-H trained with RudolfV-style self-supervision.","strengths":["Multi-institution, multi-scanner data"],"limitations":["Proprietary"],"since":2025},"route":"/technologies/atlas-aignostics/","neighbours":{"section":[{"id":"ai-computation","kind":"section","name":"AI & Computation","route":"/fronts/ai-computation/"}],"technology":[{"id":"pathology-foundation-model","kind":"technology","name":"Pathology & radiology foundation models","route":"/technologies/pathology-foundation-model/"}],"company":[{"id":"aignostics","kind":"company","name":"Aignostics","route":"/companies/aignostics/"}],"institution":[{"id":"charite","kind":"institution","name":"Charité Universitätsmedizin Berlin","route":"/institutions/charite/"},{"id":"mayo-clinic","kind":"institution","name":"Mayo Clinic","route":"/institutions/mayo-clinic/"}],"roadmap":[{"id":"ai-oncology-clinic","kind":"roadmap","name":"AI in the oncology clinic: from narrow cleared tools to multimodal decision support","route":"/roadmaps/ai-oncology-clinic/"}]}}