testmail.app alternative: GrabMail vs testmail.app

At a glance

Fourteen rows. Every testmail.app cell is what its pricing page and documentation said on 2 September 2026; every GrabMail cell is read from the constants the service enforces. “Partly” means a paid tier, a cap, or a caveat the row cannot hold — the sections below say which.

Checked 2 September 2026. testmail.app’s pricing page lists “Full API access” from Essential upward; its free plan is described as 100 emails a month with one day of retention. Custom domains appear only on the Unlimited plan. Where a figure is not published, the cell says so.

What testmail.app does well

testmail.app is what a small team builds when it wants email testing done properly and nothing else. Four things it does that GrabMail does not:

• Live queries. A GraphQL query with livequery: true waits on the server until a matching message exists, for up to a minute, then redirects to itself so the client keeps waiting. That is a polling loop you do not have to write. GrabMail’s REST API is polled; only its MCP server waits.
• Search across a run. Wildcards on subject and body, filters by tag prefix and time, and a SpamAssassin score and report on every message. A test can ask “every message to run42-* since the suite started” in one query. GrabMail lists one mailbox at a time.
• Private, scoped namespaces. Mail in a namespace is readable only with a key authorised for it, and a team can hand each project its own key. Every GrabMail mailbox is public.
• Published rate limits and soft quotas. The documentation gives exact figures — more than 10 requests a second per IP, 5 a second per paid key sustained for an hour — and paid plans allow temporary surges above the monthly allowance. Few services in this category put numbers in writing.

Where GrabMail is different

GrabMail is built for the case testmail.app’s free plan is too small for and its account model rules out: a job that needs a real inbox with nothing to sign, store or count.

• No key, no account, no namespace. testmail.app assigns you a namespace and issues a key; every call carries it. GrabMail assigns nothing: run-42@grabmail.io is an inbox the moment mail arrives, and reading it is a GET with the address in the query string.
• No monthly count. testmail.app Free stops at 100 emails a month; a suite that signs up ten users a run uses that in ten runs. GrabMail has no monthly allowance — one read a second per address, all month, on the free tier.
• Your own domain, free. At testmail.app a custom domain is on the Unlimited plan, $269 a month billed annually. Here it is one MX record at smtp.grabmail.io, and every address on the domain is a catch-all inbox — the answer to sign-up forms that refuse a testing domain.
• An MCP server for agents. testmail.app has none. GrabMail’s at https://grabmail.io/mcp gives an agent an inbox and a blocking wait_for_message of up to 25 seconds, with no key.

Price, side by side

testmail.app’s prices are per month billed annually, with a higher monthly-billing figure, and are non-refundable except at its discretion. GrabMail’s only paid plan buys access to a pool of domains kept off the disposable-mail blocklists; everything in the table above stays free.

Prices as published on 2 September 2026. testmail.app allows temporary surges above a paid allowance and asks for an upgrade if they persist. GrabMail’s pricing page has the full grid.

For developers: reading a mailbox

The whole difference in one command. This reads a GrabMail mailbox with no header, no key and no account:

$ curl -sG https://grabmail.io/api/v1/mailbox --data-urlencode "address=anything@grabmail.io" | jq '{count, alias}'

The testmail.app equivalent is a GraphQL query for inbox(namespace, tag) with a bearer key, or the JSON API with apikey, and with livequery it waits for you. It is the nicer API to write a test against — and it needs the account behind it. On GrabMail the wait is a loop with a deadline; the Playwright and Cypress guides have it, and the OTP guide covers pulling the code out.

The addressing habit carries over almost unchanged: testmail.app’s namespace.tag@inbox.testmail.app becomes tag@grabmail.io here, or tag@your-domain.com once the MX record is set. A suite that builds tags from a run id keeps doing so.

For AI agents

testmail.app publishes no MCP server and no llms.txt; an agent can call its API with the team’s key like any script. GrabMail’s server at https://grabmail.io/mcp gives an agent an inbox with no key: create_inbox, a blocking wait_for_message of up to 25 seconds, read_message.

For an agent that must sign itself up to something, the keyless server is the difference between finishing and stopping at a form. Email for AI agents covers the loop and the guard rails.

Switching from testmail.app

Both are inbound-only, so the switch is the address and the query:

1. Change the address builder. {namespace}.{tag}@inbox.testmail.app becomes {tag}@grabmail.io, or a domain of your own with its MX at smtp.grabmail.io.
2. Replace the live query with a wait loop. A GET on /api/v1/mailbox every second until count is above zero or the deadline passes. The guides have it in TypeScript and Python.
3. Remove the key. Delete the API key from CI and the GraphQL client from the lockfile. Nothing replaces them.
4. Check what you relied on. Search across tags, spam scores and 30-day retention have no equivalent here: one mailbox per call, 5 days, no scoring. Attachments still work, up to 5 MB.

Which one, for what

The honest split:

When to stay with testmail.app

Three situations where switching would cost you something real:

• Your tests carry real data. A namespace is private to its keys; every GrabMail mailbox is public, on the public domains and on yours. A test that mails a customer’s details needs the first.
• Your suite searches. If tests ask for “all messages to run42-* with subject Welcome*”, that is one query there and a loop over addresses here. Keep the query.
• You are on Essential and it is enough. $9 a month for 10,000 emails and a real API is not a cost worth engineering around. Switch when the key in CI, the monthly count or the domain price is the thing that hurts.

Questions