SMTP Relay Cost Calculator

What the comparison covers, and what it leaves out

The calculator returns the published base rate for transactional volume at the volume you enter. That number alone does not represent the bill that will actually arrive in your billing system — it is the floor. Most senders end up paying 40-60% above the published rate once dedicated IPs, email validation, deliverability suite add-ons, and overage fees are factored in (Aeroleads 2026 analysis of Mailgun pricing). The published-rate comparison is useful as a starting filter, not a procurement decision.

The dedicated infrastructure line in the calculator is the alternative most growing senders eventually consider, and the math turns favourable above roughly 500K-1M emails per month. Self-hosted PowerMTA, KumoMTA (free open source), or Postfix on a production VPS becomes cost-competitive once volume sustains warming and operational complexity is acceptable. Below that volume, the relay providers usually win on cost and deliverability — a dedicated IP that cannot generate engagement is worse than a shared pool with active reputation management.

2026 published rates and the gotchas behind them

Each provider in the calculator has a published rate and a list of cost components that surface only after you commit. The table below summarises the operational reality — the published headline plus the line items that recur in real procurement conversations.

Three practical implications. First, AWS SES is the lowest-cost line on every comparison and the highest-complexity option to operate — if your team does not include someone who can configure SES bounce handling, complaint suppression, and CloudWatch monitoring, the time cost cancels the price advantage. Second, Postmark's per-email model means the cost is predictable at low volume and grows linearly — at 1M sends, it is the most expensive option in this list, but its deliverability is consistently top-tier and that has measurable revenue value. Third, Mailgun's December 2025 Flex pricing change (doubling from $1 to $2 per 1K) caught many migration buyers off guard; the Foundation and Scale plans are now strictly cheaper than Flex above ~18K monthly emails.

How email type changes the right answer

The calculator allows you to specify transactional, marketing, or mixed because the right provider differs by use case. The volume math is the same; the operational fit is not.

Transactional sending

Receipts, password resets, account verifications, alerts. The recipient is expecting the email; engagement is high; time-to-inbox matters more than cost-per-email. Postmark is the canonical choice (per-email model, fastest delivery in independent tests). SendGrid Pro is a strong second for teams already in the Twilio ecosystem. AWS SES is the right choice for technically sophisticated teams that can build the operational tooling around it.

Marketing sending

Newsletters, promotional campaigns, lifecycle flows. The recipient may or may not engage; volume is large; cost-per-email matters more. Mailgun Scale and SendGrid Essentials/Advanced compete here. Mailgun's deliverability suite and 30-day log retention are advantages at scale; SendGrid's marketing automation is better for non-technical operators.

Mixed (transactional + marketing on same account)

The pattern most growing programmes end up in. The risk: marketing complaints damage transactional placement when both run on the same IPs. Stream isolation matters. Either use a provider that supports per-stream IP separation (SendGrid Pro, Mailgun Scale with multiple dedicated IPs) or run two separate accounts (one transactional-focused, one marketing-focused). Self-hosted infrastructure makes stream isolation trivial because additional IPs are nearly free.

Volume thresholds where each provider wins

The comparison output ranks providers by base cost at your specific volume, but the practical answer is more nuanced. Each provider has a sweet spot — a range where their pricing model and feature set produces the best total value — and the calculator's lowest-cost output may not match the provider that should actually win at that scale.