Python SDK for Vercel APIs and Vercel Functions.

For product-level behavior, limits, and platform setup, use the official Vercel docs.

Install the package:

pip install vercel

Or with uv:

uv add vercel

Configure the credentials via environment variables based on your app's needs:

• BLOB_READ_WRITE_TOKEN for Vercel Blob
• VERCEL_TOKEN for Vercel API clients such as deployments and projects
• VERCEL_PROJECT_ID and VERCEL_TEAM_ID for helpers that resolve project-scoped credentials, such as Sandbox
• VERCEL_OIDC_TOKEN for local OIDC testing. On Vercel, OIDC helpers can read the request token after you register request headers with vercel.headers.set_headers().

For local OIDC development, you can load a short-lived token dynamically with the Vercel CLI:

VERCEL_OIDC_TOKEN=$(vc project token some-project) some-command

import asyncio

from vercel.client import AsyncVercel
from vercel.oidc.credentials import get_credentials


async def main() -> None:
    credentials = get_credentials()
    vercel = AsyncVercel(access_token=credentials.token)
    deployment = await vercel.deployments.create_deployment(
        body={
            "name": "hello-python",
            "project": "hello-python",
            "target": "preview",
            "files": [
                {
                    "file": "index.html",
                    "data": "<h1>Hello from Python</h1>",
                }
            ],
        }
    )

    print(f"Deployment created: https://{deployment['url']}")


asyncio.run(main())

Import from public modules under vercel.*, such as vercel.blob, vercel.cache, vercel.headers, vercel.oidc, vercel.projects, and vercel.sandbox. Modules under vercel._internal.* are implementation details and may change without public API guarantees.

Sync counterparts are available for the main client classes and module-level helpers when you are not running an async application.

See CONTRIBUTING.md for information on development setup and process.