Buyer constraint
Cheapest production-ready Docker hosting
Not the cheapest that runs a container. The cheapest you would put in front of users.
The rule, stated before the list
Always-on, paid, public container records at a minimum 1 vCPU and 2 GB RAM. This is a cost-candidate predicate, not proof of production readiness.
A plan reaches the table below only if all of this is true of it:
- It is a serverless container plan.
- It is a paid plan, not a free tier.
- It is the always-on shape of that plan, not one that scales to zero between requests.
- The vendor publishes a monthly price for it.
- Its published size is at least 1 vCPU and 2 GB.
- A public IPv4 address is included.
- This rule finds cost candidates. The dataset cannot prove production readiness, and this page does not claim to.
Plans that qualify
| Provider | Plan | vCPU | RAM | Disk | Transfer | Price | Checked |
|---|---|---|---|---|---|---|---|
| Koyeb | always-onNot ranked: the source check is too old1 vCPU · 2 GB · no disk100 GB transfer, checked 2026-07-25 | 1 | 2 GB | none | 100 GB | $10.51 | 2026-07-25 |
| Fly.io | always-onNot ranked: the source check is too old1 vCPU · 2 GB · no disk0 GB transfer, checked 2026-07-25 | 1 | 2 GB | none | 0 GB | $11.27 | 2026-07-25 |
| Alibaba Cloud | always-onNot ranked: the source check is too old1 vCPU · 2 GB · no disk0 GB transfer, checked 2026-07-25 | 1 | 2 GB | none | 0 GB | $25.28 | 2026-07-25 |
| Amazon Web Services | always-onNot ranked: the source check is too old1 vCPU · 2 GB · no disk100 GB transfer, checked 2026-07-25 | 1 | 2 GB | none | 100 GB | $36.04 | 2026-07-25 |
Caveats and plans that just miss
These came close enough to be worth knowing about. Some meet the broad rule above but fail a narrower use-case check; the failed column names that distinction. Publishing those exclusions makes it possible to tell a deliberate caveat from a plan nobody evaluated.
What actually decides this
The dataset can identify cheap always-on container cost candidates, but it cannot prove that any record is production-ready. The executable filter requires a paid serverless-container record, an always-on scenario, at least 1 vCPU and 2 GB RAM, a known monthly total, and public ingress. Four records pass: Koyeb, Fly, Alibaba ECI, and AWS Fargate. That is the honest comparison envelope.
The current table ranks the derived single-container cost floors from the bundle. Those floors do not include a second replica, managed database, persistent disk, log retention, support, or a load balancer unless the underlying product already internalizes that function.
Free rows fail for concrete reasons. Render spins a free web service down after 15 idle minutes, uses an ephemeral filesystem, and does not allow scaling beyond one instance. Koyeb's free instance scales to zero, cannot run workers, and cannot attach volumes. A low-price IONOS VPS is a valid self-managed alternative, but that boundary transfers patching, Docker upgrades, firewalling, backups, monitoring, and recovery to the buyer.
Use this page to establish a cost floor, then run the production-readiness check. Confirm image support, restart policy, health checks, rollout and rollback behavior, secret storage, persistent data design, metrics, logs, support, and failure domains. If one instance is acceptable, say so. If two zones are required, price two instances and the routing layer. Do not label the lowest always-on row production-ready until those facts are sourced. Today the data supports a cheapest candidate, not a final production winner.
How to check this yourself
Every price here is read from the vendor's own page and carries the date we read it, most recently 2026-07-25. The rule above lists every check applied, so you can disagree with the threshold rather than guess at it. The full method is here, and the dataset is downloadable if you would rather run your own.
Compare every provider on your own workload ·All VPS plans ·Report a wrong price