DEV Community: Thiago Alvarez

Resend vs SendGrid (2026) — SendGrid Killed Its Free Tier, Now What?

Thiago Alvarez — Mon, 13 Apr 2026 19:47:12 +0000

SendGrid removed its free plan in May 2025. If your comparison article still mentions a free tier, it's lying to you. Here's what email APIs actually cost in 2026.

We Chrome-verified every number below by visiting the actual pricing pages. Not the docs. Not a cached snapshot from 2024. The real pages, in a real browser, this month.

The free tier that vanished

For years, SendGrid's free tier was the default answer to "how do I send email from my app." A hundred emails a day, forever, no credit card required. It was how most of us learned to send transactional email.

That's gone. SendGrid now offers a 60-day trial with 100 emails per day. After 60 days, you're paying $19.95/month or you're not sending email. Every tutorial, every Stack Overflow answer, every blog post that recommends SendGrid's free tier is out of date.

This matters because email is infrastructure. It's not something you choose once and revisit quarterly. You pick a provider, wire it in, and forget about it — until the pricing changes and you're scrambling.

Resend: the modern default

Resend gives you 3,000 emails per month on the free tier. There's a 100-per-day cap that can be annoying for staging environments, but for production use, 3,000 monthly emails is enough for a lot of early-stage projects.

The paid tier starts at $20/month for 50,000 emails. At 100,000 emails/month, you're looking at $90/month on the Scale plan. These numbers are nearly identical to SendGrid's — $19.95/month for 50K, $89.95/month for 100K. The price difference is negligible.

So if the cost is the same, why pick Resend? Two words: developer experience.

React Email changes everything

Resend's killer feature is first-class React Email support. If you're building with React or Next.js, you can write your email templates as React components. Same syntax, same tooling, same mental model. No more wrestling with table-based HTML from 2004.

import { Html, Button, Text } from '@react-email/components';

export default function WelcomeEmail({ name }) {
  return (
    <Html>
      <Text>Hey {name}, welcome aboard.</Text>
      <Button href="https://app.example.com">
        Get started
      </Button>
    </Html>
  );
}

That's a real email template. It renders to the gnarly inline-CSS HTML that email clients need, but you never have to look at it. You write React, Resend handles the rest.

SendGrid has nothing comparable. Their template system is a drag-and-drop editor or raw HTML. For developers who live in code, that's a step backward.

The Resend API is just cleaner

This isn't subjective. Compare the two.

Resend:

curl -X POST https://api.resend.com/emails \
  -H "Authorization: Bearer re_123" \
  -d '{"from":"you@example.com","to":"them@example.com","subject":"Hello","html":"<p>Hi</p>"}'

SendGrid's API has accumulated years of cruft. Multiple API versions, inconsistent naming conventions, nested objects for things that should be flat. It works, but it feels like it was designed by committee over a decade — because it was.

Resend's API was designed from scratch in 2023 with a clean, minimal surface. We'd rate the DX at 9 out of 10 versus SendGrid's 6 out of 10.

When SendGrid still wins

We're not going to pretend SendGrid is dead. It still has legitimate advantages.

Marketing email. Resend is transactional only — password resets, order confirmations, notifications. If you need marketing campaigns, newsletters, and transactional email in one platform, SendGrid does both. With Resend, you'd need a separate service like Mailchimp or Loops for marketing.

HIPAA compliance. SendGrid offers a Business Associate Agreement for HIPAA-covered entities. Resend doesn't. If you're in healthcare, this might be a hard requirement.

Existing integrations. If you're already on SendGrid and it's working, migration has a real cost. The API is different, webhooks are different, analytics are different. "It ain't broke" is a valid reason to stay.

The Twilio ecosystem. If you're using Twilio for SMS, voice, and video, consolidating email under the same billing and support umbrella has operational value.

The vendor health question

Resend is a YC-backed startup with $21.5M in funding and a growing developer community. They're focused, hungry, and shipping fast.

SendGrid is owned by Twilio, which has been through multiple rounds of layoffs and strategic pivots since the acquisition. Developer trust has been declining. The free tier removal is part of a broader cost-cutting pattern.

Neither is going away tomorrow, but the trajectories are pointing in different directions.

The decision

Pick Resend if you're starting a new project. You want the best developer experience available. You're building with React or Next.js and want React Email templates. You only need transactional email. You want a free tier that actually exists.

Pick SendGrid if you need transactional and marketing email in one platform. You need HIPAA compliance with a BAA. You're already integrated and it's working fine. You're in the Twilio ecosystem and want consolidated billing.

Pick neither if you're sending fewer than 300 emails a month. Use your hosting provider's built-in SMTP or Amazon SES at $0.10 per thousand. Don't add a dependency you don't need.

One more thing

We didn't just verify Resend and SendGrid. We Chrome-verified pricing for 74 cloud services across email, auth, databases, payments, hosting, and more. Every number on auxiliar.ai comes from a real browser visit — not cached documentation, not an LLM hallucination.

Get it in your terminal:

npx auxiliar

Or add to Claude Code as an MCP server:

claude mcp add auxiliar -- npx auxiliar-mcp

Every comparison. Every price. Verified.

Neon vs Supabase (2026) — Database or Backend? The Real Tradeoffs

Thiago Alvarez — Mon, 13 Apr 2026 19:46:46 +0000

Neon got acquired by Databricks for $1B. Supabase hit 1M databases. PlanetScale killed its free tier. The Postgres landscape shifted — here's what actually matters in 2026.

We Chrome-verified every pricing page for this comparison. No cached docs, no guessing, no "last time we checked." Just numbers from a real browser, as of this month.

This isn't a database comparison

The biggest mistake in every Neon-vs-Supabase thread on Reddit is framing it as a database comparison. It isn't. Neon is a database. Supabase is a backend platform that includes a database. Comparing them head-to-head on Postgres features misses the point entirely.

Neon gives you PostgreSQL. That's it. Serverless scaling, database branching, a built-in connection pooler, and a serverless driver that works on edge runtimes. It does one thing and does it well.

Supabase gives you PostgreSQL plus authentication, file storage, realtime subscriptions, edge functions, and Row Level Security integrated with its auth system. It's closer to Firebase than it is to a managed database.

If you only need a database, Supabase means you're paying for a platform you'll use 20% of. If you need auth, storage, and realtime alongside your database, Neon means you're gluing together three separate services yourself.

The pricing reality

Both services have free tiers, and both have catches.

Neon's free tier gives you 0.5 GB of storage. The catch is 500ms cold starts. Your first query after idle spins up the compute, and that half-second delay is noticeable in production. To get always-on compute with no cold starts, you need the Scale plan at $69/month. The Launch plan at $19/month still has cold starts — it just gives you more storage and compute.

Supabase's free tier gives you 500 MB of storage. The catch is automatic pausing — your project goes to sleep after one week of inactivity. The Pro plan at $25/month keeps it running and removes the pause. And crucially, that $25/month gives you always-on compute with no cold starts, which is $44/month cheaper than Neon's equivalent.

Read that again. If cold starts matter to you — and they probably do — Supabase gives you always-on Postgres for $25/month versus Neon's $69/month. That's a significant gap.

Where Neon wins: branching

Neon's killer feature is database branching. You can create a full copy of your production database — schema and data — in seconds, use it for CI/CD or preview deployments, and throw it away when you're done. It's like git branches for your database.

If you're running a team with pull request previews, each PR can get its own database branch with real production data. That's transformative for testing. Supabase doesn't have anything equivalent.

For solo developers or small teams that don't use preview deployments, this feature won't matter much. For teams with mature CI/CD pipelines, it's a legitimate reason to choose Neon despite the higher price for always-on compute.

Where Supabase wins: everything else

Supabase's auth integrates with Row Level Security at the database level. Your security policies live in SQL, enforced by Postgres itself. That's elegant and hard to replicate with separate services.

Realtime subscriptions let you build collaborative features without standing up a WebSocket server. File storage comes with CDN and image transformations. Edge functions handle serverless compute. It's a lot of infrastructure you don't have to set up or maintain.

The DX is also excellent — we'd rate it 9 out of 10. The dashboard, the documentation, the client libraries — Supabase has invested heavily in developer experience, and it shows.

After the Databricks acquisition

Neon's $1B acquisition by Databricks raises a question: what happens to the developer-facing product? Databricks is an enterprise data company. The optimistic read is that Neon gets more resources and deeper integration with enterprise tooling. The pessimistic read is that the indie-developer-friendly product slowly pivots toward enterprise customers.

Supabase, by contrast, is a YC-backed startup with $116M in funding and a clear commitment to the developer community. Their trajectory is more predictable, for better or worse.

The decision framework

Pick Neon if you only need a database. You want database branching for CI/CD. You're deploying on edge runtimes where Neon's serverless driver shines. You want serverless scale-to-zero for projects with unpredictable traffic.

Pick Supabase if you want auth, database, storage, and realtime in one platform. You're building a full-stack app and want the fastest setup. You want always-on Postgres without paying $69/month. You'll actually use the bundled services — otherwise you're just overpaying for a database.

Pick neither if you need maximum control. Spin up your own Postgres on a $5/month VPS. It's more work, but it's the cheapest and most flexible option at any scale.

One more thing

We didn't just verify Neon and Supabase. We Chrome-verified pricing for 74 cloud services across databases, auth, email, payments, hosting, and more. Every number on auxiliar.ai comes from a real browser visit — not cached documentation, not an LLM hallucination.

Get it in your terminal:

npx auxiliar

Or add to Claude Code as an MCP server:

claude mcp add auxiliar -- npx auxiliar-mcp

Every comparison. Every price. Verified.

Clerk vs Better Auth (2026) — We Verified Every Price So You Don't Have To

Thiago Alvarez — Mon, 13 Apr 2026 19:46:18 +0000

Every auth comparison you've read is already wrong. Clerk changed pricing. Better Auth added organizations. Auth.js went maintenance mode — and then the Auth.js team joined Better Auth in September 2025. We visited every pricing page in a real browser to get you the truth.

We built auxiliar.ai specifically because of this problem. Cloud service pricing changes constantly, and stale blog posts keep ranking on Google long after the numbers are wrong. So we Chrome-verified everything below. Here's what we found.

The real cost of Clerk

Clerk is the fastest way to add auth to a Next.js app. Ten minutes, pre-built UI components, social providers, organizations — it all works out of the box. The developer experience is genuinely a 9 out of 10.

But Clerk's pricing model has a gotcha that bites at scale. The free tier covers 10,000 monthly active users. That's generous for a side project. Once you cross that line, you're paying $25/month plus $0.02 per additional MAU.

Let's do the math that matters. At 100,000 MAU, your Clerk bill is roughly $2,025/month. That's not a typo. Two thousand dollars a month for authentication. For a lot of indie developers and early-stage startups, that's a hard pill to swallow — especially when the alternative is free.

The other catch: Clerk stores your user data on their servers in the US. There's no EU data residency option. If you're building for European users, you're relying on the Data Privacy Framework, which is legally fragile and has already been challenged in court twice.

Better Auth: the Auth.js successor nobody expected

Better Auth is open source and self-hosted. Your data lives in your database, in whatever region you choose. The cost at any scale is zero — beyond your own hosting, which you're already paying for.

When the Auth.js team joined Better Auth in September 2025, it signaled a real shift. Auth.js was the go-to for Next.js authentication, and its spiritual successor is now Better Auth. That matters for long-term community support and ecosystem integration.

The DX is solid but not as polished — we'd rate it a 7 out of 10. The biggest gap is pre-built UI components. Clerk gives you drop-in sign-up and login forms that look professional. With Better Auth, you're building those yourself. Budget 1-3 days for UI work on top of the 20-minute core setup.

Better Auth supports 40+ social providers via plugins, has native SvelteKit support, and gives you full control over the auth flow. No vendor lock-in. No per-user pricing surprises. No data residency headaches.

The decision that actually matters

Forget feature matrices. The real decision comes down to two questions.

How fast do you need to ship? If you're building a prototype, entering a hackathon, or validating an idea this weekend, Clerk gets you there in 10 minutes with production-quality UI. That speed has real value. You can always migrate later — though "later" has a way of becoming "never" once your user table lives on someone else's servers.

What does your cost curve look like? If your app could realistically hit 50K-100K users, the math changes dramatically. At 100K MAU, Clerk costs $2,025/month. Better Auth costs whatever your Postgres instance costs, which is probably $25-50/month. Over a year, that's a $24,000 difference. That's a developer's salary in some markets.

Pick Clerk if...

You need auth working today, not next week. Your team doesn't want to build login and signup UI from scratch. You're under 10,000 MAU and cost isn't a concern yet. You don't need EU data residency. You value speed-to-ship above everything else.

Pick Better Auth if...

You need GDPR-compliant EU data residency. You're cost-sensitive at scale — anything above 50K MAU. You want to own your auth stack with zero vendor lock-in. You're building with SvelteKit, where Better Auth has native first-class support. You care about the Auth.js lineage and want the community that comes with it.

One more thing

We didn't just verify Clerk and Better Auth. We Chrome-verified pricing for 74 cloud services across auth, databases, email, payments, hosting, and more. Every number on auxiliar.ai comes from a real browser visit — not cached documentation, not an LLM hallucination.

Get it in your terminal:

npx auxiliar

Or add to Claude Code as an MCP server:

claude mcp add auxiliar -- npx auxiliar-mcp

Every comparison. Every price. Verified.