Get started with serverless sandboxes

Run a command in a serverless sandbox and print its output. No cluster setup is required.

Run your first sandbox

Choose W&B if you don’t have a CoreWeave console account or prefer W&B billing. Choose CoreWeave to use your CoreWeave account.

W&B
CoreWeave

Get your W&B API key. If you don’t have an account, sign up through the same link. Then set the key in your terminal:

export WANDB_API_KEY="[WANDB-API-KEY]"

Python
TypeScript

With uv installed, save this as hello.py:

hello.py

from cwsandbox import AuthStrategy, Sandbox

with Sandbox.run(
    auth=AuthStrategy.WANDB,
    resources={"cpu": "1", "memory": "1Gi"},
) as sandbox:
    print(sandbox.exec(["echo", "Hello!"]).result().stdout)

Run it with Python 3.11 and the sandbox client. The command installs them if needed:

uv run --python 3.11 --with 'cwsandbox[wandb]>=1.10.0' hello.py

With Node.js 22 or later installed, install the client:

npm install @coreweave/cwsandbox

Save this as hello.mts:

hello.mts

import { createSandboxClientFromEnv } from "@coreweave/cwsandbox/wandb";

const client = createSandboxClientFromEnv();
const result = await client.withSandbox(
  async (sandbox) => sandbox.commands.run(["echo", "Hello!"]),
  { resources: { cpu: "1", memory: "1Gi" } },
);
console.log(result.stdout);

Run it with tsx:

npx --yes --package=tsx tsx hello.mts

These examples use a key from wandb.ai.

Ask your administrator to create or update an Access Policy granting you the Sandbox User role. See Create an access policy for the steps.Create an API access token on the Tokens page, copy its Token Secret, and set it in your terminal:

export CWSANDBOX_API_KEY="[API-ACCESS-TOKEN]"

Python
TypeScript

With uv installed, save this as hello.py:

hello.py

from cwsandbox import AuthStrategy, Sandbox

with Sandbox.run(
    auth=AuthStrategy.COREWEAVE_API_KEY,
    resources={"cpu": "1", "memory": "1Gi"},
) as sandbox:
    print(sandbox.exec(["echo", "Hello!"]).result().stdout)

Run it with Python 3.11 and the sandbox client. The command installs them if needed:

uv run --python 3.11 --with 'cwsandbox>=1.10.0' hello.py

With Node.js 22 or later installed, install the client:

npm install @coreweave/cwsandbox

Save this as hello.mts:

hello.mts

import { createSandboxClientFromEnv } from "@coreweave/cwsandbox/node";

const client = createSandboxClientFromEnv();
const result = await client.withSandbox(
  async (sandbox) => sandbox.commands.run(["echo", "Hello!"]),
  { resources: { cpu: "1", memory: "1Gi" } },
);
console.log(result.stdout);

Run it with tsx:

npx --yes --package=tsx tsx hello.mts

The example prints the following output and stops the sandbox when it finishes:

Output

Hello!

The sandbox has 1 CPU and 1 GiB of memory. Serverless sets resource limits equal to requests. Change resources to choose its CPU and memory. For concurrent quotas, rate limits on create requests, and quota errors, see Limits and quotas.

CoreWeave Serverless sandboxes are in public preview.

For account requirements and token setup, see Choose and get credentials. If your organization uses W&B Dedicated Cloud, follow Use sandboxes with W&B Dedicated Cloud.

Next steps

Continue with the guide for your workload:

Use your own compute: enable a runner on your cluster and run a sandbox there.
Sandbox placement and spillover: compare serverless and CKS capacity and configure fallback.
Run coding agents: connect a coding agent to a sandbox.
GPU sandboxes: request GPU access and run a GPU workload.
Python client: execute commands, work with files, and configure sandboxes.

​Run your first sandbox

​Next steps

Run your first sandbox

Next steps