The family

Three sizes of compute, and storage that stacks.

ON Edge S, ON Edge M, ON Edge X and three stacked ON Store units
Desk

ON Edge S

1 PFLOP AI128 GB memory

One lab, one bench. Runs segmentation, pathology encoders and a local language model side by side.

Bench

ON Edge M

2 PFLOPS AI256 GB memory

A shared box for a group, with room for larger models and several people at once.

Rack

ON Edge X

20 PFLOPS AI748 GB memory

Department-scale compute for the largest models and batch runs across whole studies.

Storage

ON Store

120 TB per unitStack to scale

Storage that sits next to the compute. Add a unit as your images and data grow.

Specifications are design targets and may change before release.

Setup

From box to first result in minutes.

  1. Unbox and place it on the bench.
  2. Connect power and network.
  3. Pair it with OMĒOS under Hardware, Compute.
  4. Run a model from Studio. Here, nuclei detection on an H&E slide.
ON Edge S front panel showing OMĒOS starting up ON Edge S back panel with power and network cables ON Edge S running on the lab bench
Grows with your lab

Add capacity when you need it.

Start with one box. Add ON Store units as your data grows, and an ON Edge X when the models do. Everyone on the team uses them from OMĒOS.

Why on site

Your data stays where it was made.

Data stays put

Slides, movies and sequencing runs are large. Analyse them where they are, not after an upload.

Models fit on a desk

Segmentation, pathology and protein models run comfortably on an ON Edge S.

One workspace

ON Edge appears in OMĒOS like any other computer, next to your laptop, and soon ON Cloud.