Get started

1. Join Anyscale

• Contact your organization's admin and request an invitation.
• Use the email invitation link to complete your registration and create your Anyscale account.

First user to Anyscale?

If you are the first Anyscale user on your team and doesn't have an Anyscale organization, follow Get started for admins to create and set up your Anyscale organization.

2. Try it out

Once you create your account, follow the quickstart tutorial in the console to learn about the platform.

Beyond the quickstart tutorials, there are many examples to help you learn Ray and expedite your development process. Choose from deploying and fine-tuning large language models, scaling batch inference, serving Stable Diffusion, and much more.

3. Set up your environment (optional)

To programmatically interact with Anyscale outside the console and from your terminal, you need to install the Anyscale CLI.

AWS

Step 1: Install the Anyscale CLI and Python client package

pip install -U anyscale

Step 2: Authenticate the CLI

To access Anyscale's services, you need to obtain a token that verifies your identity. Running the following command fetches and updates the token in the local credential file ~/.anyscale/credentials.json:

anyscale login

You can also manually generate your API token through the console, and set it in the ANYSCALE_CLI_TOKEN environment variable.

GCP

Step 1: Install the Anyscale CLI and Python client package

pip install -U "anyscale[gcp]"

Step 2: Authenticate the CLI

To access Anyscale's services, you need to obtain a token that verifies your identity. Running the following command fetches and updates the token in the local credential file ~/.anyscale/credentials.json:

anyscale login

You can also manually generate your API token through the console, and set it in the ANYSCALE_CLI_TOKEN environment variable.

What's next?

➡️ Thinking in Ray and Anyscale - Learn about the high-level thought process of building a scalable Ray app on Anyscale.