Mailgun vs Resend vs SendGrid: pick your transactional email service in 2026
Mailgun vs Resend vs SendGrid in 2026: which transactional email service is worth your money for a small Indonesian dev team on a tight budget.
Three founders I know in BSD City area have gone through all three of these services. Two of them migrated away from SendGrid. Here is what actually matters when you pick a transactional email service for a small team shipping real software.
What we are comparing
Mailgun — owned by Sinch since 2021. The original developer email API. Strong deliverability controls, good log retention, HTTP API plus SMTP. Pricing has crept up since the Sinch acquisition.
Resend — launched 2023 by ex-WorkOS engineers. Built React Email support from day one. Modern API, excellent DX, fast-growing. Younger company.
SendGrid — owned by Twilio. The oldest and most enterprise of the three. Deep analytics dashboard, rich event webhooks, dedicated IP options. Also the most expensive at scale and the most complex to configure correctly.
All three handle the same core job: deliver OTP emails, password resets, receipts, and notification emails reliably. The difference is cost, developer experience, and how much control you have over deliverability.
Comparison table
| Dimension | Mailgun | Resend | SendGrid |
|---|---|---|---|
| Free tier | 100 emails/day (card required) | 3,000/month (no card) | 100/day, 3,000/month |
| Paid entry point | ~USD 15/month (Flex) | USD 20/month (Pro) | USD 19.95/month (Essentials) |
| 100K emails/month cost | ~USD 75 | ~USD 89 | ~USD 89.95 |
| API quality | Good, REST + SMTP | Excellent, modern SDK | Good, but complex |
| React Email support | Via SMTP only | First-class, native | Via SMTP, partial |
| Log retention (free) | 1 day | 3 days | 3 days |
| Dedicated IP | Yes (from Flex) | Not yet available | Yes (extra cost) |
| Webhooks | Yes | Yes | Yes, more event types |
| Dashboard UX | Functional, dated | Clean, modern | Feature-heavy, overwhelming |
| SDK ecosystem | Node, Python, Ruby, PHP | Node, Python, PHP | Node, Python, Ruby, Go, PHP |
Mailgun
Mailgun was the go-to choice for developer email for years, and it still earns that trust in specific scenarios. Their deliverability controls are genuinely granular — you can manage sending domains, set up inbound routing, configure suppression lists, and monitor bounce rates with more precision than SendGrid’s shared-IP environment.
The pricing model post-Sinch acquisition is honest: you pay per email on Flex, no monthly seat fees. For a startup sending 50,000 emails/month that does not want to pay a flat subscription, Mailgun Flex can be cheaper than SendGrid Essentials.
The weakness is the 1-day log retention on the free plan. When you are debugging a production issue at 2am — your OTP emails are not arriving for users in the Pasar Modern area of BSD — and your logs have already expired, that is a painful limitation. Paid plans extend this to 30 days, which is reasonable.
Mailgun’s dashboard feels like 2017 software. Functional, not pleasant. If your team includes junior developers who need to inspect email events without CLI access, the UX friction is real.
Resend
Resend is the service I recommend first to most small teams in 2026. The developer experience is the best of the three — not by a small margin.
The API is clean. The Node SDK is typed properly. React Email integration means you write email templates as React components, preview them in a browser, and ship them without maintaining a mess of HTML strings or Handlebars templates. For a Next.js or Remix team, this is a significant reduction in email template maintenance overhead.
// Resend with React Email — this is what "good DX" looks like
import { Resend } from 'resend';
import { OtpEmail } from './emails/OtpEmail'; // your React component
const resend = new Resend(process.env.RESEND_API_KEY);
await resend.emails.send({
from: '[email protected]',
to: user.email,
subject: 'Your OTP code',
react: <OtpEmail code={otp} userName={user.name} />,
});
The 3,000/month free tier without a credit card is genuinely useful for validating a product before charging users. Most Indonesian B2C apps testing product-market fit will stay under that limit in early stages.
The notable gap: no dedicated IP yet. If you need IP warming for high-volume sends or stricter deliverability SLAs, you cannot get a dedicated IP from Resend as of September 2026. For most SaaS teams under 500K emails/month, this does not matter. For fintech or high-volume notification systems, it does.
SendGrid
SendGrid does more than the other two, and that breadth is both the value and the problem. The dashboard has marketing email tools, contact lists, segmentation, A/B testing — features your small team does not need and will navigate around constantly.
Deliverability on the shared IP pool can be inconsistent in the first weeks. This is a known issue — not unique to SendGrid, but more noticeable here because so many low-quality senders share those IPs. Moving to a dedicated IP fixes this but adds setup time and cost.
Where SendGrid wins is enterprise situations: large teams that need role-based access, deep event webhooks for compliance logging, dedicated IP warming workflows, and integrations with CRMs. For a team of 15+ with a dedicated DevOps person, SendGrid’s feature set makes sense.
For a four-person team in a startup co-working space off Jalan Pahlawan Seribu shipping their first SaaS, SendGrid is solving problems you do not have yet.
Verdict
Pick Resend if: your team writes React or TypeScript, you want the fastest path from zero to working transactional email, and you are under 500K emails/month. This covers most Indonesian SaaS and side projects right now.
Pick Mailgun if: you need dedicated IP control without paying SendGrid prices, you send high volume and want per-email pricing (not flat subscription), or you need SMTP relay for a non-Node stack. Also the better choice if your ops team is comfortable with a dated but functional dashboard.
Pick SendGrid if: your company is past Series A, has compliance requirements, needs enterprise-level event logging, or your infrastructure team has already standardized on Twilio for SMS and wants one vendor for communications.
Do not start with SendGrid just because it is the most well-known name. The shared IP deliverability issues on the entry plan will cost you debugging hours that a smaller, more focused service avoids entirely.
Transactional email is not exciting infrastructure. Pick the one that gets out of your way fastest, and move on to building the actual product.
Ditulis oleh Asti Larasati