frameworkdjango · fastapi
Detected or declared
Django, FastAPI and Flask are recognised from the repository; anything else runs from a start command or a Dockerfile.
/python-hosting.html
Python hosting
Django, FastAPI and Flask, without the ops rota.
Deploy Python web services and Celery workers from Git, with managed Postgres alongside them and memory that stays allocated — useful when your application loads models or holds long-running work.
Python 3.10–3.13 · pip, Poetry and uv · Gunicorn and Uvicorn · Celery workers · no cold starts
What Python services get
workercelery resident
Celery that stays up
Background workers are separate always-on services with their own CPU and memory, not opportunistic processes.
memoryup to 8 GB
Room for models
Applications that load an embedding model or hold caches in memory keep that allocation for the whole month at a fixed price.
migrateon release
Migrations in the build
Run collectstatic and migrate as part of the release step so a deploy either completes fully or rolls back.
cronscheduled jobs
Scheduled tasks
Cron-style jobs run in the same environment as your app, with logs in the same place as your deploys.
dbpostgres 16
Postgres next door
Managed Postgres in the same region with the connection string injected — the usual Django or SQLAlchemy setup, minus the admin.
What a Python service costs
Django applications with a worker typically run one Pro or Pro Plus box for the web service and a smaller box for the worker. AI or data workloads that hold models in memory usually want Pro Plus or Scale.
See all plans and what is included →Deploying a Python app
Connect and detect
Point at the repository; requirements.txt, pyproject.toml or a Dockerfile is picked up.
connect github → your-org/api
Set the start command
Gunicorn for WSGI, Uvicorn for ASGI — edited in the panel, not in a config file.
gunicorn app.wsgi -w 3 -b :8080
Add the worker
Deploy the Celery worker as a second service from the same repository.
celery -A app worker -l info
Python questions
Python 3.10 through 3.13. Declare it in pyproject.toml, a runtime.txt or your Dockerfile and the build follows it.
More operational detail is on the full FAQ page.
Put one Python service on it.
Deploy a web service and a worker from the same repository, and see how the fixed monthly price compares.
Keep your existing service online while you evaluate DebutDeploy.