Helm7 compared with Vercel and Supabase
Vercel and Supabase are the default way to ship a Next.js app with a database behind it, and the pairing works. This page is about what it costs to keep two vendors in the loop, and what you give up by collapsing them into one.
Both products are good at what they do. Most of what follows is arithmetic, not criticism.
- Vercel and Supabase, worked example
- $195
- Helm7, before usage
- $12 minimum
- Rows where they are better
- 3
Three people, one application in production, one custom domain, point-in-time recovery on.
How Vercel and Supabase is put together
Vercel runs the front end. It is very good at that: a large edge network, fast builds, preview deployments on every pull request, and the people who write Next.js also run it.
Supabase runs the state. Postgres, auth, storage, realtime and edge functions, with a client library that a lot of application code is already written against.
The seam is where the cost sits. Two dashboards, two bills, two incident pages, two sets of environment variables to keep in step, and no single place that can tell you what one application costs to run.
What Vercel and Supabase costs
Published list prices, read from their own pricing pages on 22 September 2026. If one of these has moved, tell us and we will fix it.
Their list prices
- Vercel Pro
- $20 per seat, per month
- Vercel data transfer over the included allowance
- $0.15 per GB
- Supabase Pro
- $25 per month
- Supabase point-in-time recovery
- $100 per month for 7 days of retention
- Supabase custom domain
- $10 per month, per project
Charged per person on the team, whether or not they deploy.
Per organisation.
Priced per 7-day window, so 14 days is $200.
Worked through
Three people, one application in production, one custom domain, point-in-time recovery on.
- Vercel Pro, three seats
- $60
- Supabase Pro
- $25
- Supabase PITR, 7 days
- $100
- Supabase custom domain
- $10
$195 a month before any usage, from two invoices.
Try it with your own team
Vercel
Monthly invoice
- Pro, 3 seats$60
Supabase
Monthly invoice
- Pro$25
- Point-in-time recovery, 7 days$100
- Custom domain$10
Two invoices, before usage$195
Helm7
Monthly invoice
- Pro minimum$29
- 3 peopleSeats are not charged$0.00
- Point-in-time recoveryIncluded on Pro$0.00
- Custom domainIncluded$0.00
One invoice, before usage$29
Vercel and Supabase figures are their published list prices, read on 22 September 2026. Usage is extra on every side of this; on Helm7 it draws down against the minimum instead of being added to it.
What Helm7 costs
$12 a month runs one application in production. Each price is a minimum monthly spend. Usage draws down against it, and you only pay more if you use more.
No per-seat charge, and no charge for an agent key. Before you build anything, ask what it would cost: cost_estimate prices a shape without creating it, and every change returns a plan with its monthly cost on it.
Side by side
The rows where they win are marked. A comparison table with no losses in it is an advert, and you would be right not to trust the rest of the page.
| Capability | Helm7 | Vercel and Supabase |
|---|---|---|
| Seats | Not charged. Two people or thirty agents costs the same. | Vercel is $20 per seat, per month. |
| Bill | One invoice for compute, Postgres, storage and domains. | Two invoices, neither of which knows about the other. |
| Cost before you build | cost_estimate prices a shape without creating it. Every change dry-runs to a costed plan. | Usage dashboards after the fact. No pre-flight estimate. |
| Postgres | Postgres 18 with pgvector, pooled, provisioned as part of the Product. | Postgres with a large extension set and a mature dashboard. |
| Point-in-time recovery | Included from the Pro plan. | $100 a month per 7 days of retention, on top of Pro. |
| Custom domains | Attached to a Product. No per-domain surcharge on the database side. | Free on Vercel; $10 a month per project on Supabase. |
| Auth and storage | GoTrue-compatible auth and Supabase-Storage-compatible buckets, so supabase-js works against a changed URL and key. | The originals, with a bigger surface and far more written about them.They are better here |
| Realtime subscriptions | Not built. There is no Helm7 equivalent of Supabase Realtime. | Realtime is a first-class product with years behind it.They are better here |
| Edge network | Two EU regions. No US region yet. | A global edge network with points of presence close to most users.They are better here |
| Long-running work | Workers and cron services with no execution time limit. | Vercel functions have a maximum duration. Long work needs a queue or another host. |
| Agent access | Every operation exists as an MCP tool, generated from the same definition as the REST route. Keys carry spend caps and expiries. | REST and management APIs exist. There is no plan-then-apply flow and no per-key spend cap. |
How a move goes
- 1
Point Helm7 at the repository
product_create takes a repo URL, inspects it and proposes a shape. Run it with dry_run: true first and read what it plans to build.
- 2
Move the database
Create the Postgres instance, then restore a Supabase dump into it with pg_restore. Both sides are ordinary Postgres, so this is a normal dump and load, not a conversion.
- 3
Repoint supabase-js
Auth is GoTrue-compatible and storage is Supabase Storage compatible. In most applications this is a URL and a key in the environment, not a code change. Anything using Realtime or Edge Functions is the exception and needs rewriting.
- 4
Copy the environment variables
secret_set per variable, per environment. Values are write-only afterwards: nothing in the API reads them back.
- 5
Deploy and compare
Deploy to a Helm7 domain and run both in parallel behind the old DNS until you are satisfied.
- 6
Move the domain
domain_attach returns the DNS record to create. The certificate issues once the record resolves, usually within a minute.
A small application moves in an afternoon. One that leans on Realtime, Edge Functions or Vercel-specific middleware takes longer, and the honest answer is that we cannot tell you how much longer without seeing it.
When you should not choose Helm7
There are cases where the right answer is to stay where you are. Here they are, so you find them now rather than three weeks into a migration.
Your users are in North America or Asia
Helm7 runs in two EU regions today: Falkenstein (eu-central) and Helsinki (eu-north). There is no US region yet. A request from San Francisco crosses the Atlantic. Vercel will be faster for those users, and no amount of caching hides a cold request.
You depend on Supabase Realtime
There is no equivalent in Helm7. If your product pushes live changes to clients through Supabase channels, moving means rebuilding that part.
You need the Vercel-specific parts of Next.js the day they ship
Vercel writes Next.js and runs it first. Helm7 supports Next.js well, but new platform features land there before they land here.
The bill is already small
A solo developer on Vercel Hobby and Supabase Free pays nothing. Consolidating costs money. Do it when the seat charges and the PITR line start to hurt, not before.
You want a long operating record
Helm7 is new. There is no large community, no third-party plugin ecosystem, and no long public operating record to point at.
If one of these applies to you, say so and we will tell you straight whether it is worth waiting. [email protected].
Try it on the trial first
One Product for 72 hours, no card, full API. Point it at the repository you are thinking of moving and read the plan it returns. That takes ten minutes and tells you more than this page does.