Achieve Cloud Hardware Independence
Say goodbye to hardware-specific packaging, tuning, and APIs that lock you into a hardware vendor. OctoML allows your trained model to run in production – anywhere.
Deploy to the cloud – in hours, not months
The OctoML Platform optimizes and containerizes ML models into a flexible, CI/CD friendly deployment package that can be repackaged and redeployed in a few clicks. We automate the pain of managing a deployment stack, version management, and manually tuning models for different hardware targets.
Get hardware portability
The OctoML Platform sits across all three major clouds and major chip provider options in these clouds. Select your current targets and also additional hardware to explore how your optimized model performs across all options.
You indicate your business SLA constraints, and the Platform recommends the best targets for reducing costs, maximizing throughput, and minimizing latency.
Choose from flexible packaging options
We provide your choice of packages that are immediately deployable to your hardware targets, including: Python, Docker, shared library with C API, and more. OctoML handles the dependency management by packaging all the requirements to deploy your model. We make it easier to migrate to cost saving or better performing hardware.
