Jupyter Notebook to Serverless ML Endpoint.

CloudKit is our solution for serverless ML inference, all with a single predict function in your Jupyter Notebook.

with just one more cell

Add a predict function, consume some input data, run ML inference, then return output data. CloudKit handles all the nasty details.

NatML Logo

run ML models from anywhere

CloudKit runs inference requests on GPU-enabled servers, allowing you to leverage ML from literally anything with an internet connection.

NatML Logo


CloudKit is designed to be extremely simple. Pay-as-you-go, per second spent on inference.


Quickly build a prototype.

$0.0005 /second

Join the Waitlist

What's included:

  • Run predictions on Intel CPUs.
  • Create unlimited public endpoints.


Deploy serverless ML to production.

$0.001 /second

Join the Waitlist

Everything in Free, plus:

  • Accelerate predictions with Nvidia GPUs.
  • Create private endpoints with limited access.


Enhanced integration and support.


Contact Sales

Everything in Team, plus:

  • Dedicated servers with no cold starts.
  • Host CloudKit on-prem.
  • Priority support, including Slack Connect with our engineering team.