better authvsclerk

what you're actually comparing

Better Auth is an authentication library, not a service. It's MIT-licensed, runs inside your TypeScript application, and writes users into your own database. Everything is plugin-based: social providers (40+ of them), passkeys, MFA, and an organization plugin that generates the schema, server APIs and client SDK for orgs, members, invitations and role-based access control. In late 2025 Auth.js — formerly NextAuth.js — joined forces with it, which consolidated a fragmented corner of the ecosystem.

Clerk is a hosted authentication service. You get drop-in sign-up and sign-in components that look professionally designed out of the box, a user management dashboard, session handling, organizations, and the operational burden of all of it sitting with Clerk rather than you.

The real distinction isn't features — both cover the modern checklist. It's that one is code you run and the other is a service you rent.

pricing, honestly

Better Auth is free. MIT license, no tiers, no per-user component. Your cost is the database you were already paying for.

Clerk got substantially more generous on February 5, 2026, raising its free allotment from 10,000 to 50,000 monthly retained users. Pro is $25/month for production features and includes those 50,000 MRU, then $0.02 per user in the 50,001–100,000 band, with published rates all the way past a million. It's worth appreciating that Clerk bills on retained users rather than active ones — people who sign up, look around once and vanish don't count, which makes the allowance larger in practice than the number suggests.

So the honest version: below 50,000 users, both are effectively free and cost is not a reason to choose. Above it, the curve diverges hard — at 100,000 monthly users Clerk runs on the order of $2,000/month while Better Auth still costs whatever your Postgres instance costs. Over a year that gap is real money, and it arrives exactly when you're least able to stop and re-platform your authentication.

edge: better auth at scale, neither below 50k.

where better auth wins

You own the users. They're rows in your database, in a region you picked, queryable with the same SQL as everything else. Joining users to your own data is a join, not an API call and a sync job.

Data residency. Clerk is US-only as of 2026 with no EU residency option. If you're a European business with GDPR obligations, that's frequently the end of the conversation before features or price come up at all.

No pricing cliff. The bill never scales with signups. A viral week is an infrastructure problem, not an invoice problem.

No vendor risk. MIT-licensed code running in your app can't be repriced, sunset, or acquired out from under you.

Plugin depth without a plan tier. Passkeys, MFA, organizations, and 40+ social providers are plugins, not features gated behind an upgrade.

where clerk wins

You ship today. Drop in the components and you have a sign-up flow that looks better than what most teams would build in a week. Better Auth gives you the primitives; you still design the screens.

The dashboard. Clerk's user management UI — finding a user, impersonating them, revoking a session, debugging why someone can't log in — is real operational value that you'd otherwise build yourself the first time support asks for it.

Someone else is on call. Auth is the one system where an outage locks every user out of everything. With Clerk, fixing that at 3am is somebody else's job.

Session and token handling is solved. Refresh, rotation, revocation, edge cases across subdomains — all handled, all tested at a scale your app hasn't reached yet.

Compliance paperwork. SOC 2 reports you can forward to a customer's security team, without producing them yourself.

things to know before you pick

Auth is the most expensive thing to migrate later. Every other dependency can be swapped incrementally. Auth touches every session, every user record, and every password hash you can't move. Choose as if you won't get to choose again.

Clerk's free tier change matters more than it looks. At 50,000 MRU, a great many products will never pay Clerk anything. If you're picking Better Auth purely on cost for a product that will never cross that line, you're optimizing a bill that would have been zero.

Better Auth means you own the incidents too. No per-user bill also means no one to escalate to. Budget for the operational work rather than pretending it doesn't exist.

Check residency before you build. This catches European teams late and painfully. If EU data residency is a requirement, resolve it in week one, not during a security review.