← Home

Free tools

One of These Free Postgres Tiers Deletes Your Data on a Timer

Neon, Supabase and Render all say "free". One sleeps your compute, one pauses your project, and one expires the database 30 days after you create it and then deletes it.

Comparing free Postgres tiers by storage is comparing the wrong number. All three of these are within a factor of two of each other on storage, and they differ completely on the thing that actually matters.

The question is what happens when you look away

Render’s is the one that catches people, and it catches them because the word “free” does not distinguish between “free forever with limits” and “free trial”. The clock is not on idleness — it starts when you create the database. Use it every single day and it still expires on day 30.

That is a perfectly reasonable product. It is just not a free tier, and the docs are the only place that says so.

Put your own workload in

Note that Render never reports “fits”, no matter how small you make the numbers. That is deliberate: a tier that deletes your data on a schedule should not return the word “fits” for a 50 MB database, because “fits” gets read as “safe to put real data in”.

The limit that actually runs out is egress

Everyone compares storage. Storage is rarely the binding constraint on a side project — 500 MB is a lot of rows.

Egress is the one that bites. Both Neon and Supabase include 5 GB of monthly transfer. A dashboard that pulls a 200 KB list on every page load hits that at about 25,000 page loads a month, which is a successful side project rather than an unusual one. And egress is charged on the way out, so a chatty client costs you more than a large database does.

Cold starts are a design constraint, not a footnote

Neon’s five-minute suspend cannot be turned off on the free plan. If your project is a cron job, a dashboard you open twice a week, or a demo, that is exactly right and you should take it.

If there is a human waiting on an HTTP request that hits a cold database, you have moved a several-hundred-millisecond wake-up into your request path, and no amount of application caching hides the first one.

What to actually check before you pick

1. What happens to the data when you ignore it for a month. This is the only question with an irreversible answer, so it goes first.
2. Egress, not storage. Multiply your average response size by realistic monthly traffic before you look at gigabytes of disk.
3. Whether a cold start is acceptable in your hot path. If it is, mandatory scale-to-zero is a feature; if not, it disqualifies the tier regardless of the other numbers.

And check the current values yourself. The ones above were read off each provider’s own pages on 6 August 2026, and free tiers move faster than anything else on a pricing page.