Python Tor Hosting — Django, Flask & FastAPI on .onion

AnubizHost's Python Tor hosting supports every major Python web framework. Django deployments include Gunicorn as the WSGI server with Nginx reverse proxy, PostgreSQL for the database, and Redis for caching and Celery task queues. We configure Django settings for production security, including proper SECRET_KEY management, ALLOWED_HOSTS set to your .onion address, and secure cookie settings.

Flask and FastAPI applications are equally well-supported. Flask apps run behind Gunicorn with gevent or eventlet workers for concurrent request handling. FastAPI applications use Uvicorn with ASGI for native async support, delivering excellent performance for API-heavy workloads. Both frameworks benefit from our Nginx front-end for static file serving and connection management.

We also support less common frameworks like Pyramid, Tornado, Sanic, and Starlette. If it runs on Python, it runs on our Tor hosting. Provide a requirements.txt or Pipfile and a WSGI/ASGI entry point, and we configure the rest. Virtual environments are used for every deployment to prevent dependency conflicts.

Deploy Python applications via our Tor-accessible Git hosting. Push your code, and our deployment system creates a fresh virtual environment, installs dependencies from your requirements file, runs database migrations, collects static files, and restarts Gunicorn. All deployment steps are logged and visible in your control panel for debugging.

We support pyproject.toml, setup.cfg, requirements.txt, Pipfile, and Poetry lock files for dependency management. Pip packages are downloaded through Tor to prevent clearnet exposure. We maintain a PyPI mirror cache that significantly speeds up common package installations — popular packages like Django, Flask, and their dependencies install in seconds from the local cache.

For reproducible deployments, we recommend using pip-compile to generate pinned requirements files with exact version numbers and hashes. This ensures that every deployment installs identical dependencies regardless of when it runs. Our deployment system verifies package hashes automatically when they are present, catching any tampering during download.

Most Python web applications need background task processing. Our hosting includes Celery with Redis as the message broker, configured to run alongside your web application. Define your tasks in your Django or Flask app, and they execute asynchronously in Celery worker processes. Beat scheduling handles periodic tasks like database maintenance, report generation, and data aggregation.

For simpler scheduling needs, we configure systemd timers or cron jobs that run Python scripts at specified intervals. These are useful for tasks that do not need the full Celery infrastructure — database backups, log processing, health checks, or automated content updates. Timer definitions are version-controlled alongside your application code.

Long-running processes like data scrapers (over Tor), blockchain node synchronization, or machine learning training jobs can run as separate systemd services with their own process management and logging. PM2 alternative solutions like Supervisor are also available for managing multiple Python processes with automatic restart and resource limits.

Python's dominance in data science and machine learning makes our Tor hosting an excellent platform for privacy-preserving analytics and AI services. Install NumPy, pandas, scikit-learn, PyTorch, and TensorFlow from our cached PyPI mirror. Serve machine learning models behind a FastAPI endpoint accessible only through your .onion address.

Jupyter notebooks can be hosted as a Tor hidden service for private data exploration and collaboration. Access your notebook server through Tor Browser with token-based authentication. This setup is valuable for research teams handling sensitive data who need a collaborative analysis environment without exposing their work to the clearnet.

For GPU-accelerated workloads, our dedicated server plans offer NVIDIA GPU options with CUDA support. Train models on private data sets without sending your data to cloud providers. Combined with Tor's anonymity properties, this enables truly private machine learning — neither the training data nor the model's existence is visible to outside observers.