New to Machine Learning? Check out

Enterprise Runs on Spell

Start-ups and big businesses alike use Spell to accelerate discovery and manage their end-to-end machine learning pipeline.

Request a demo


Try all the latest hardware and DL frameworks without managing versions or signing up for additional providers.

The fastest hardware

nvidia logo



google logo

Tensor Processing Unit (TPU)

any language framework

mxnet logo
numpy logo
Run details

Experiment & Collaborate

See your whole team’s world and progress in one dashboard. View active runs, compare experiments, and see long-term results in our web console.

Jupyter is the best way to investigate, test, and share work among your engineers. And Spell brings hosted and GPU-enabled Jupyter notebooks to an organization.


The spell hyper command provides built in support for hyperparameter optimization and gets you to better models 10x faster.

Facet chart
Model servers


Easily keep track of models and results, and deploy them to Kubernetes/Kubeflow with one click.


Our Workflow API and Metrics API give you control over the entire lifecycle of your projects. Set up custom pipelines to ingest data and run experiments, and dashboards and charts to track the performance of your code.

Automation infographic

Contact Us

Connect with our team to learn how Spell can accelerate your business.

Contact Us