Beam is currently in private beta. If you’re reading this, you’re one of the first people to try it.
📦 Setup remote development environments in code
Configure your runtime in Python - tell us how many GPUs you need and which libraries you want installed, and Beam will spawn a remote environment for you.
🛰 Develop locally on remote hardware
You can write and run your code locally - except when you enter your shell, your code will run on Beam instead of your local machine.
🚀 Deploy apps as serverless functions
Deploy your apps as serverless REST APIs, scheduled cron jobs, or webhooks - all in just four lines of Python.
What can you do with Beam?
Deploy ML models on serverless runtimes. Your app will scale automatically with traffic and spin down when idle.
Develop on your local IDE, while running on remote GPUs
Instantly jump from hacking on a script to serving an API in production
Save money on your cloud bill. Beam is serverless and charges simple usage-based, per-second pricing
Tutorial: Deploying Stable Diffusion on GPU
Get started with an example
Stable Diffusion on GPU
Deploy Stable Diffusion using a GPU
Scraping a website and running the results through an ML model
Huggingface REST API
Deploying a pre-trained ML model as a REST API
Apps. Each project in Beam is called an app. When you first start Beam, you’ll be prompted to define your environment through the
Beam.App()method in the SDK.
Triggers. Triggers are actions that can invoke your Beam apps. For example, a REST API trigger allows your Beam app to be invoked via a REST API. A webhook trigger will allow your Beam app to be invoked asyncronously using a webhook, and so on.
Outputs. Outputs are file paths that can be used to save files created when your functions are run.