PythonAnywhere review

PythonAnywhere is a hosted Python development environment and web-hosting service, not a Python package, local IDE or server you install in your own infrastructure. Its browser interface provides a code editor, Python consoles, server-side storage and tools for deploying web applications. The vendor says you can develop and host code without installing software or managing your own server.

For a web stack, PythonAnywhere sits between your application and the underlying machines: you supply a Python WSGI app, and the service handles the hosting environment. Quickstart installers cover Django, web2py, Flask and Bottle; the vendor says it can also host other WSGI frameworks. Scheduled tasks and browser-accessible consoles cover scripts and interactive work alongside web traffic.

That managed model reduces server setup and patching work, but it also means your app depends on PythonAnywhere’s environment, account limits and supported services. It is a fit for small sites, prototypes, coursework and simple scripts—not a drop-in replacement for unrestricted Linux infrastructure.

The main integration surface is the hosted browser environment: write Python in the web editor, open browser-based consoles, upload or download files, and deploy a WSGI web app. The vendor lists quickstart installers for Django, web2py, Flask and Bottle, plus support for other WSGI frameworks. It also lists Git, Mercurial and Subversion project access, scheduled tasks, and preconfigured Python versions and libraries.

The vendor doesn't publish a public REST API or SDK list on its homepage. It presents PythonAnywhere as a browser-based service rather than an API-first platform, so teams automating deployment or account operations should check the current documentation for supported interfaces and confirm SDK support before building a workflow around them. The homepage points users to Help pages, guides, FAQs and forums; it does not name a CLI or a specific language SDK.

For teaching, teachers can view student files and consoles and distribute files. That makes the service usable as a shared learning environment, but it is not a general deployment pipeline or infrastructure-as-code interface.

Beginner is free and includes one web app, up to two consoles, 100 CPU-seconds per day, 512MB of private storage and restricted outbound internet. That is enough to learn or host a small project if its scripts stay within the daily CPU allowance and it does not need unrestricted external requests.

Developer costs $10 per month, or $8.33 per month with annual billing. It includes one web app, three web workers, 5,000 CPU-seconds per day, 5GB of private storage and 20 scheduled tasks. Custom is listed at $10–$500 per month, with up to 20 web apps, up to 100,000 CPU-seconds per day, configurable workers and disk, and a choice of always-on tasks. Paid plans have a one-month minimum and can be canceled anytime; annual subscriptions pay for 10 months per year.

Additional CPU seconds are available, but the price is not published. Optional private PostgreSQL and storage costs vary by configuration. The pricing page does not publish a free trial.

PythonAnywhere is presented as a hosted service, not as a server users can install in their own cloud or data center. The vendor doesn't publish an open-source license or a self-hosting option, so treat the platform as proprietary and hosted-only unless the vendor confirms otherwise.

You can access files and work with Git, Mercurial or Subversion projects, which gives you ways to keep application code in a repository and move files. That is not the same as a complete account export: plan settings, scheduled tasks, web-worker configuration and any managed database need to be recreated or moved separately. The vendor doesn't publish a one-click migration or complete data-export process.

A typical WSGI app can be moved to another Python host, but expect to review Python versions, installed packages, environment settings, database compatibility, filesystem paths and task scheduling. Keeping the app and its deployment notes in version control reduces the work if you need to leave.

The production fit depends on workload shape. The Developer plan lists 5,000 CPU-seconds per day, three web workers and 20 scheduled tasks; Custom offers configurable capacity up to 100,000 CPU-seconds per day. The vendor says a $10 Developer account can support a 10,000-hit-per-day website, but that is a vendor claim, not a capacity guarantee for every app. Check CPU use, worker needs, storage and outbound access against your own traffic and dependencies.

Outbound internet restrictions are especially important: the free plan explicitly restricts it, and one Reddit thread reports that the restriction prevented calls to an external embeddings API. Another Flask discussion reports trouble using pgvector with PythonAnywhere’s PostgreSQL version and says threading and Celery were not supported for that user’s workload. These are developer reports, not universal compatibility statements.

A Django Reddit thread reports recurring outages; treat that as a reason to verify current status history and support expectations, not as an uptime measurement. Ask the vendor about backups, incident response, security controls and compliance needs before hosting critical or regulated data.