FlatRun vs Railway
FlatRun is an open source, self-hosted alternative to Railway. Push to deploy, preview pull requests, roll back, and share variables between projects, with your apps running as plain Docker Compose on your own server.
Open source, MIT licensed · Runs on any Linux server with Docker
Hosted platform
Railway and FlatRun share a model: a project holds services, every push becomes a deployment, pull requests get their own environment, and variables can point at other services. The difference is where it runs and how you pay for it.
Railway runs your services on its own infrastructure and bills a plan fee plus CPU, memory, storage and egress by the second. FlatRun runs on a server you rent or own, so the bill is the server, however many projects it holds. Each project stays an ordinary directory with a Compose file, so it keeps running if you stop using FlatRun.
Side by side
| FlatRun | Railway | |
|---|---|---|
| Where apps run | Your own servers | Railway's infrastructure |
| License | Open source, MIT | Proprietary, hosted |
| Pricing | Free. You pay for the server | Plan fee (Hobby $5, Pro $20 per workspace each month) plus usage billed per second |
| Compute cost | Fixed by the server you choose | About $20 per vCPU and $10 per GB of memory each month |
| Deploy on push | GitHub, GitLab and generic webhooks | Yes |
| Builds | Dockerfile, Railpack, or a Compose file | Dockerfile or Railpack |
| Pull request previews | An environment per pull request, removed when it closes | An environment per pull request, removed when it is merged or closed |
| Rollback | Deploy any of the last 50 deploys again | Redeploy an earlier deployment |
| Variable references | ${{ project.KEY }} and ${{ secret.NAME }} | ${{ service.VAR }} and ${{ shared.VAR }} |
| Secrets | Sealed, rotated on a schedule, or read from Infisical, Vault, Doppler or 1Password | Sealed variables |
| Sleep when idle | Per project, after a set number of idle minutes | Serverless sleeping after inactivity |
| Data | Folders beside the Compose file | Railway volumes |
| Change preview | Plan and apply before a change runs | Staged changes before deploy |
Same ideas
| Railway | FlatRun | What it is on your server |
|---|---|---|
| Project | Project | A directory with a Compose file |
| Service | Service | A service in that Compose file |
| Environment | Environment | A sibling project that follows its own branch, such as shop-staging |
| Deployment | Deployment | One deploy, with its commit, trigger, phase and logs |
| Variables | Variables | Written to the project's .env.flatrun when it deploys |
| Shared variables | Workspace secrets and references | A value one project reads from another with ${{ project.KEY }} |
| Volume | Data folder | A bind mount beside the Compose file |
| Template | Template | A ready-made Compose project |
Switching
Run the installer on an Ubuntu or Debian server. It sets up the agent, the dashboard and the proxy.
Create a project from the same GitHub or GitLab repository and branch. A Dockerfile or Railpack build works as it did on Railway.
Paste the service variables, seal the secrets, and restore database dumps into the project's database service.
Add the domain to the project and update DNS. FlatRun issues the certificate.
For apps built from a repository with a Dockerfile or Railpack, the workflow is the same: connect the repository, push, preview pull requests, roll back. Railway's managed regions and per-second scaling have no equivalent; FlatRun scales on the servers you give it.
FlatRun is free and open source. The cost is the server: a few dollars a month for a small VPS, and the same price whether it holds one project or twenty.
The syntax is close. Railway writes ${{ service.VAR }}; FlatRun writes ${{ project.KEY }} for another project's variable and ${{ secret.NAME }} for a sealed secret.
Railway details come from Railway pricing , Railway variables , Railway environments and Railway deployments . FlatRun details are in the documentation.
More comparisons