Mystic AI FAQs
Cut down cold start times by up to 90% with Mystic Turbo Registry, the custom Docker registry and containerd adapter that loads ML models up to 15x faster.
FAQs of Mystic AI
What is Mystic AI?
Mystic AI is a platform that allows you to easily deploy and scale machine learning models. We offer two ways to deploy models: in your own cloud (AWS, GCP, Azure) or in our shared cloud.
How does Mystic AI work?
Mystic AI allows you to deploy your ML models to your own cloud, or to our shared cluster of GPUs, in just a few steps. You can use our open-source Python library to wrap your AI pipeline and then deploy it with a single command. Once your model is deployed, you can access it through RESTful APIs, our CLI tool, or our dashboard.
What are the benefits of using Mystic AI?
Mystic AI offers a number of benefits, including:
- Cost optimization: You can run your models on spot instances and use your own cloud credits.
- Fast inference: Our platform uses vLLM, TensorRT, TGI, or any other inference engine you choose, and we have a custom container registry that reduces cold starts.
- Simple developer experience: We provide a managed platform that removes the complexities of building and maintaining your own ML platform.
- Scalability: Our platform automatically scales up and down GPUs depending on the number of API calls your models receive.
How does Mystic AI compare to other ML deployment platforms?
Mystic AI is different from other ML deployment platforms because it is designed for both performance and cost optimization. You can run your models on spot instances, which are much cheaper than on-demand instances, and you can use our custom container registry to reduce cold starts. We also have a simple developer experience, so you don't need to be a Kubernetes expert to use our platform.
How to use Mystic AI
- Begin by creating an account on the Mystic AI platform. This grants access to the AI model deployment tools.
- Utilize the Mystic AI Python library and associated APIs to package your machine learning pipeline. This simplifies deployment.
- Employ the
pipeline container pushcommand to upload your packaged pipeline to the Mystic AI registry. This prepares it for deployment. - Following the upload, a new version of your pipeline is automatically deployed to your cloud environment. Mystic handles scaling.
- Use the provided RESTful APIs, CLI, or dashboard to interact with and manage your deployed AI model. Monitor performance metrics.
