Choosing an SMTP provider for your newsletter in 2026: SES vs Postmark vs Mailgun vs Resend

Whatever you send email from (your app, your CMS, a self-hosted Listmonk box) is almost certainly not a mail server. It writes the message and hands it off to something else to deliver. That something else is an SMTP provider, and here's the uncomfortable bit: inbox providers judge that provider's reputation, not yours. You can send a perfectly clean, opted-in, correctly authenticated newsletter and still land in spam because the provider you relayed through has IPs sitting on a blocklist. Their problem becomes your problem.

That makes the choice a bigger deal than it looks, and once you've spent months building a domain reputation on top of a provider, switching is a pain. We've run most of these in production, so this is a working read on the four you'll be choosing between in 2026: Amazon SES, Postmark, Mailgun, and Resend. I'm not going to crown one. What you're sending matters (a password reset and a weekly newsletter are different animals, and providers treat them differently), and so does how much of the surrounding work you want to do yourself.

The short answer

If you send in real volume and have engineering time, Amazon SES is by far the cheapest (~$0.10 per 1,000) but hands you a firehose and no tooling. If deliverability is the product and you send transactional mail, Postmark is the most curated, but it's US-only, which rules it out if you need EU data residency. Mailgun sits in the middle on price and flexibility and has an EU region. Resend has the best developer experience and a generous free tier, but the shortest track record. And for a newsletter specifically, none of the four is a complete answer. Sending is one job out of about ten, which is the real point of this post.

Amazon SES

SES is dirt cheap. Around $0.10 per 1,000 emails, which is roughly ten to fifteen times less than the premium options. At a million emails a month that stops being a detail and becomes real money saved.

The problem is getting in. New SES accounts are stuck in a sandbox that only sends to addresses you've verified, which is worthless for a real list. Getting out means applying for production access, and AWS turns those down more than people expect, especially if your use case looks like a newsletter platform or "sending on behalf of customers." The reasons they give are vague, the appeal is slow, and there's nobody to call. Plain transactional sending usually gets approved. Anything that smells like bulk or multi-tenant, keep a backup plan ready.

And SES hands you a firehose and not much else. Bounce handling, reputation dashboards, deliverability tooling: you build all of it yourself out of other AWS pieces. Great if you've got the engineering time and want the control. A slog if you don't. On the plus side you pick your region, and SES has several in the EU (Ireland, Frankfurt, Paris among them), so if you want mail leaving from Europe instead of Virginia you can. It's just one more thing on your plate rather than something handled for you.

Postmark

Postmark's pricing is three paid tiers that all include 10,000 emails a month: Basic at $15, Pro at $16.50, Platform at $18, plus a free tier capped at 100 emails a month. What changes across the tiers is the overage rate once you pass 10,000: roughly $1.80 per 1,000 on Basic, dropping to about $1.20 on Platform. No annual billing, so the monthly figure is the figure.

You're paying for curation. Postmark stays strictly transactional and keeps broadcast mail on a separate stream, so a marketing send can't tank the deliverability of your receipts. It polices its shared IP pools hard and kicks out bad actors, which means you get to ride other people's good behaviour instead of being punished for their bad behaviour. The dashboards are the best in the group, per-message detail is clear, and support is a real human replying in hours.

Two trade-offs. First, Postmark won't do everything, on purpose. It sends transactional and broadcast mail, and it will not run your list, your segments, or your signup forms. The strictness is the whole product. Second, and this matters for a lot of European senders: Postmark is US-only. All account data, content, and logs are stored in the United States, and the company has said it has no plans to add EU servers. If you need your subscriber data to stay in the EU, that's a hard stop regardless of how good the deliverability is.

Mailgun

Mailgun sits in the middle of everything. Developer-focused, flexible, volume pricing that lands between SES on the floor and Postmark on the ceiling. Paid plans start around $15/mo for 10,000 emails, with overage rates from roughly $1.10 to $1.80 per 1,000 depending on tier. The API's mature, the docs are good, deliverability is fine.

It's always taken more of a "sure, we'll send it" line than Postmark's strict door policy. If your sending is legitimate but doesn't fit a neat transactional box, that flexibility is worth a lot. The cost is that the pools aren't policed as tightly, so at the cheaper tiers your deliverability rides on who else is in there with you, and getting support is harder than with Postmark. Unlike Postmark, Mailgun runs a real EU region (data centre in Germany): you pick it per domain, and your mail and customer data stay in Europe, which makes it one of the more GDPR-friendly options here.

Mailgun rewards knowing what you're doing and punishes not knowing. If you want more room than SES gives you without paying Postmark prices, and you don't mind owning a bit more of the deliverability work, it's a solid pick.

Resend

Resend is the new one, and it's climbed fast for good reason. The developer experience is the best here by a distance: clean API, sane SDKs, docs that don't fight you, and you're sending in minutes instead of days. The free tier is generous, 3,000 emails a month, and paid plans start around $20/mo, with per-1,000 costs (~$0.90) that fall as volume climbs. That generous entry is a big part of why so many startups now reach for it first. It also offers an EU sending region (Ireland), though note the nuance: your mail can leave from the EU, but account data and logs are still stored in the US, so it's not full EU data residency.

The catch is simply that it's young. Shorter track record than the incumbents, and track record is the thing that takes years to earn and matters most at 2am when something's broken. For transactional and simple broadcast it's very good. If you're putting a large, mission-critical sending operation on it, part of that bet is on a history that's still being written. Plenty of senders are fine making that bet. Just know you're making it.

But a newsletter is more than sending

Here's what the comparison quietly skips. For a newsletter, not one of these four is a full solution, because sending is one job out of maybe ten, and it happens to be the one they all do.

The best SMTP provider on earth only puts mail on the wire. Everything that decides whether the newsletter actually works stays with you:

• The list itself: where it lives, how it's segmented, the signup forms feeding it.
• Double opt-in, so the people on the list actually asked to be.
• A working unsubscribe flow, including the one-click List-Unsubscribe header that the 2024 bulk-sender rules made mandatory, not optional.
• Bounce and complaint handling, wired to the provider's feedback and acted on every time, so you stop mailing dead addresses and anyone who hit "spam."
• SPF, DKIM, and DMARC set up right on your domain and kept that way.
• List hygiene, the boring ongoing job of cutting dead and disengaged addresses.
• And under all of it, building sending reputation slowly and then guarding it, which takes months and can be wrecked in a weekend.

Pick a provider and you've done step one. The relay is real and it matters, but it's the wire, not the newsletter. The other nine are yours, and they're mostly the parts that quietly rot your deliverability while you're not looking. There's a deeper version of this problem (who's actually on the IPs your mail goes out from), but even the shortlist above makes the shape clear. They sell you the sending. The newsletter is everything wrapped around it.

Why we run our own SMTP

We'll be honest about our own history here, because it explains the whole product. In our early days Yellaro relayed through Postmark, and Postmark was good at its job: clean deliverability, great dashboards, real support. We're not writing this to knock it. The problem wasn't Postmark. The problem was the model.

Postmark, Mailgun and Resend all price per email. Yellaro prices per subscriber, flat, with unlimited sending: you pay for the size of your list, then email it as often as you like. Those two models are mathematically incompatible, and the arithmetic isn't close.

Take a Grow customer: €19 a month, a list of up to 5,000 subscribers. Send weekly and that's about 20,000 emails a month. On Postmark's per-1,000 rate (~$1.20–1.80) that's roughly $24–36 a month in sending alone, before the app, the hosting, the support, or anything else. The plan is underwater on the relay bill by itself. Push to a bigger list on our €79 tier (up to 50,000 subscribers, so ~200,000+ emails a month at a weekly cadence) and premium per-email pricing runs into the hundreds of dollars a month against a €79 price. There's no version of that math that works.

SES is the one exception on price. At ~$0.10 per 1,000, 200,000 emails is about $20. But SES is the firehose with no list management, no reputation isolation, and the approval risk described above. Cheap sending isn't the same as a solved newsletter.

So we did the expensive thing and built our own SMTP relay, and we isolate each customer's reputation rather than pooling everyone together. That's what lets the pricing be flat and the sending unlimited: when your cost isn't metered per email, we're not quietly hoping you send less. The trade-off is that we now own deliverability end to end, which is exactly the work most people don't want. That's rather the point. (If flat, per-subscriber pricing is new to you, we wrote it up separately.)

FAQ

Which is cheapest? Amazon SES, comfortably: around $0.10 per 1,000 emails, versus roughly $1.00–1.80 per 1,000 for Postmark and Mailgun. The catch is that SES gives you a raw pipe and no deliverability tooling, so the low price buys engineering work.

Which is best for a newsletter? None of them on their own. They only handle sending, which is about one-tenth of running a newsletter. If you want to pick one anyway, Mailgun and Resend are more comfortable with broadcast mail than strictly-transactional Postmark. A flat-rate platform that runs sending and everything around it (list, opt-in, unsubscribe, reputation) is usually the better fit than a bare SMTP provider.

Does Postmark have an EU region? No. Postmark stores all data in the US and has said it has no plans to add EU servers. If you need EU data residency, look at Mailgun (full EU region in Germany) or SES (EU regions like Ireland and Frankfurt); Resend sends from the EU but still stores account data in the US.

SES vs Postmark: which should I use? SES if you have high volume and engineering time and want the lowest possible cost (~$0.10 per 1,000). Postmark if deliverability and clear reporting matter more than price, you're sending transactional mail, and you don't need EU data residency. They sit at opposite ends: cheap-and-DIY versus curated-and-managed.

Where this leaves you

If you like running infrastructure, the choice above is a real decision and any of the four can be right, depending on your volume, your appetite for approval risk, your EU-residency needs, and how much of the surrounding work you want to keep. Pick on the merits.

Yellaro exists because we did all of it ourselves (the sending and every one of those other nine steps) and decided that most people who just want to send a newsletter shouldn't have to. We run the sending and everything around it, flat pricing by list size with unlimited sending, hosted in the EU, so the only thing left for you to decide is what to write. If you're coming from a self-hosted setup, that's the gap our managed option fills too. If that beats becoming a part-time deliverability engineer, see how it works.