SMTP relay vs email API
Pick the interface that preserves delivery state.
Compare an SMTP relay with an email API for transactional mail. See where each fits, what durable acceptance requires, and how Transmitly handles the sending path.
An SMTP relay accepts a message for onward delivery. An email API can return a durable message ID, per-recipient state, idempotency and signed delivery events. Transmitly now offers both interfaces: an API-first path and authenticated SMTP submission with the same durable acceptance transaction.
Where an SMTP relay fits
SMTP is useful when an existing application already speaks it and cannot easily make an HTTPS request. It is also a common migration path from an in-house mail server. The tradeoff is that SMTP acceptance alone does not give you a structured message object or an idempotency key unless the relay adds one.
Where an API gives you more
An HTTP API can commit the message, deliveries, idempotency key and outbox event before returning 202. It can return one message ID for a batch and expose per-recipient outcomes later. That is harder to model over plain SMTP, where the response is usually a queue ID or a bare 250.
- Idempotent retries with conflict detection.
- Per-recipient delivery state and event webhooks.
- Domain and stream scoping in the request.
- A stable message ID for support and audit.
Transmitly’s position
Transmitly is API-first and also exposes authenticated SMTP submission for applications that already speak SMTP. Connect to mail.transmitly.io on port 587 with STARTTLS or port 465 with implicit TLS, then authenticate with an API key. The username can be “apikey” and the password is the full key.
Submission is scoped, not a general-purpose relay. Every sender must be on a domain verified in the authenticated workspace, the From header must match the envelope sender, and each transaction is quota-checked and rate-limited. The server returns 250 only after the durable message, delivery and outbox transaction commits. Mail still leaves through the owned KumoMTA spool with DKIM signing, suppression checks and the same webhook events as the API.
- Ports 587 (STARTTLS required before AUTH) and 465 (implicit TLS).
- 50 MiB per message, 50 recipients, 20 attachments and 40 MiB of attachments.
- Standard MIME, inline CID images, CC/BCC envelopes and the synthetic @transmitly.test addresses.
- No open relay: unauthenticated MAIL FROM is refused and sender domains are checked against the workspace.
Authentication does not change
Either interface still needs a verified sending domain, aligned SPF and DKIM, and a DMARC policy. The relay or API is the handoff; authentication is what receivers evaluate. Transmitly shows the DNS records and verifies them before the domain becomes active.
Questions
Common questions
+Do you offer a public SMTP relay?
No. Transmitly offers authenticated SMTP submission for its own customers on mail.transmitly.io. It is not an open relay; every sender must be on a verified domain in the authenticated workspace.
+Can I migrate from an SMTP relay to the API?
Yes. You can also keep SMTP and move to the API later. Start with a verified subdomain, run a small flow in parallel, then move volume once acceptance, deferral and complaint rates look stable.
+Does SMTP or API change deliverability?
The handoff does not decide inbox placement. Domain authentication, recipient engagement, sending history and content do. What changes is how much delivery state you can observe and retry safely.
+How do I authenticate SMTP submission?
Use an API key with the send or admin scope as the SMTP password. STARTTLS on port 587 is required before AUTH; port 465 uses implicit TLS. The username can be apikey or the key prefix.
+What response should I expect?
The API returns 202 only after the durable transaction commits. Delivery to MX is a later event and is reported separately.
Related
See what the remote MX actually said.
Start free, verify a domain and send a test. Acceptance and delivery evidence stay separate.