EXAONEPath

EXAONEPath is a pathology foundation model introduced by LG AI Research, with paper arXiv 2408.00380 and a 1.0 release in August 2024. The 86-million-parameter model was pretrained on 285,153,903 patches extracted from 34,795 whole slide images, targeting tile-level representation learning for histopathology.

The vendor reports strong performance relative to the number of slides and parameters used. EXAONEPath 1.0 is distributed as a checkpoint via the project's GitHub releases and is also packaged in the NVIDIA MONAI model zoo. Later versions (Path 1.5 and Path 2.0) added multi-omics training and whole-slide analysis, but this entry covers the open 1.0 patch-level release. Check the repository for the applicable license before use.