Neon vs Turso
Two managed database options for database. When each fits, what it costs, who moves from one to the other, and what makers who chose it say.
Which fits you
- Your traffic is bursty or you run many preview deploys, and idle hours shouldn't cost much
Use it whenYou want every pull request's preview deploy to get its own copy of the database.
Trade-offCold starts after idle periods add latency to the first query, and heavy always-on workloads can cost more than a fixed server.
- SQLite at the edge, or one database per customer.
Use it whenEach customer should get their own isolated database, or reads need to happen close to users worldwide.
Trade-offSQLite semantics — one writer at a time, fewer extensions — suit read-heavy apps better than write-heavy ones.
At a glance
| Used by | 95 makers' products · 16 open-source projects | 13 makers' products · 33 open-source projects |
|---|---|---|
| Cost at default usagedatabase size 8 GB, data transfer 50 GB, document or row reads 10 million reads, document or row writes 2 million writes | $22/mo Launch | $5.99/mo Developer |
| Moved to it on GitHubpull requests since Oct 2024 | 7 from Turso | 7 from Neon |
| Downloads | 3.7M/wk+35% vs npm | 2.2M/wk+102% vs npm |
| Pricing | Free tier; usage-based paid plans. · paid from $0.106 per CU-hour | Free tier; paid plans by databases and usage. · paid from $4.99/mo |
| Free tier | Yes | Yes |
| Open source | Yes · self-hostable | Yes · self-hostable |
Cost as you grow
At 0.5 GB Turso costs less ($0 vs $20); from about 100 GB Neon does ($67 vs $67); from about 500 GB Turso does ($287 vs $457).
Who moves from one to the other
Public pull requests on GitHub since Oct 2024 whose title says "Neon to Turso" or the reverse — real code changes, by developers in general rather than makers only.
- feat(api): move the database from Turso to Neon Postgres (JEF-342)mankatcheung/trakwyn · 2026-09-18
- Port DB from SQLite/Turso to Postgres on Neon (#96, #97, #98)zachradtka/nextopp · 2026-05-27
- fix: migrate hosted DB from libSQL/Turso to Neon Postgresjoerickson/Wallet-Scanner · 2026-04-25
- feat: migrate database from Turso to Neon Postgresrlfarman/rej-studio · 2026-04-05
- feat(db): migrate from turso to neonkapishdima/shadcn-rss · 2026-01-19
- Migrate from Turso to Neonmetaideas/init · 2025-05-15
- Migrate from Turso to neontechulus/manage · 2025-04-05
- Migrate from Neon PostgreSQL to Turso SQLiteharshi79/vibebin · 2026-08-29
- feat: migrate production database from Neon to Tursottizze/digital-buddhism · 2026-08-29
- feat(repo): migrate database from Neon Postgres to Turso (libSQL)johinsDev/loyalty-app · 2026-05-27
- Migrate database from Supabase/Neon to Turso (libSQL + Drizzle)heaventree/flowbase-productive-ai-app · 2026-05-21
- Claude/migrate neon to turso e ms motobipch/CubeFeed · 2026-05-13
- Claude/migrate neon to turso e ms motobipch/CubeFeed · 2026-05-13
- feat(database): migrate database from Neon to Tursophantanhoancuong/wordest · 2026-03-09
What makers say
Makers on using it for database, from Product Hunt and Starter Story interviews, each linked to the source. Products with a page of their own and fuller notes first.
Serverless Postgres with branching is a game changer for solo builders. Scale to zero means no idle costs, and the connection pooling just works. Perfect for an early-stage product.
Neon has been great for managing our database. The branching workflow and serverless Postgres made development smooth and allowed me to iterate quickly while building UseAgents.
We use Neon for Superset's Postgres database and database branches when developing schema changes. It is part of the shared backend behind our desktop and mobile apps.
We wanted to build application that has single tenant databases for each organization. Turso handles that super nicely. In addition the ability to have embedded replicas is killer feature for building local RAG.
We love Turso because it combines the simplicity of SQLite with the scalability we need for production. It works seamlessly for edge and serverless applications, which makes it perfect for the kind of projects we build.
Turso is my database! Turso made it really easy to switch from local development to hosted on the cloud.
Loved and watch-outs
Themes that recur in makers' words and Hacker News comments, each linked to what it summarises.
- Postgres scales to zero when idle, so early products and staging databases cost almost nothing until traffic arrives. PHHN
- Branching copies the production database in seconds, so migrations and preview environments are tested against real data. PHPH 2HNHN 2
- The serverless HTTP and WebSocket driver fits Cloudflare Workers and Next.js routes without connection-pool trouble. PHHN
- Separating compute from storage costs latency and performance compared with Postgres on local NVMe. HNHN 2
- Parts of the stack, like autoscaling and some tooling, are not supported or openly licensed for self-hosting. HNHN 2
- With no built-in caching, crawler traffic can burn through the data transfer allowance quickly. HN
- It keeps SQLite's simplicity while scaling for edge and serverless apps, and moving from local development to hosted is easy. PHHN
- A database per tenant and embedded replicas make single-tenant designs and local-first or offline apps straightforward. PHPH 2HN
- The Rust rewrite adds async I/O, concurrent writes, encryption, change data capture and vector search while staying SQLite file-compatible. HNHN 2HN 3HN 4
- The Rust rewrite is still beta, with panics, data-corrupting bugs and thin test coverage that undermine trust for production. HNHN 2HN 3HN 4
- The split between libSQL and the new Turso packages is confusing, and libSQL showed memory leaks on large write jobs. HNHN 2HN 3
- Compatibility with SQLite is incomplete; for example, another process cannot read a WAL database it has open. HNHN 2

