MoonCo — an elastic GPU cloud for AGI.
❯docs.moonco.onedocumentation and access · business@moonco.io
MoonCo is an elastic GPU cloud for AGI.
We run our own GPUs and rent them on demand by the GPU-hour. Scale up for a training run, release the capacity when it finishes, and pay for the hours you used. The models we serve are also available through an API, metered by tokens.
- cloud
- elastic GPU
- hardware
- our own
- compute
- on demand, per GPU-hour
- inference
- API, per token
moonco-v1, the first model on MoonCo
moonco‑v1 is our first model, built for clinical reasoning: our own reasoning head, trained with reinforcement learning on an H‑JEPA architecture, over a selectable open-weights foundation.
- type
- clinical reasoning
- context
- 1M
- served
- API, metered by tokens
It runs where the institution keeps its data, in a private cloud or on-premises, and retains nothing by default. We do not train on customer data.
ReMedQA closed-form accuracy
Closed-form accuracy across seven equivalent versions of each question.
Benchmark results validated by Harvard Medical School.
moonco-v1 API price per million tokens
Even the best healthcare AI, ours included, is not yet capable of replacing clinical judgment. The model is built to be checked by a clinician, not to stand in for one.