Turso

Turso brings SQLite into production as a hosted, replicated database, using libSQL - an open-source fork of SQLite - so that a database engine designed for a single file on a single machine can serve an application distributed across the world. The idea behind it is that most applications do not have a database problem, they have a latency problem. A conventional setup puts one Postgres instance in one region, and every user outside that region pays the round trip on every query. SQLite is extraordinarily fast, but it was never built to be reached over a network. Turso's answer is to replicate small SQLite databases to locations near your users and read from the closest copy, with writes funnelled to a primary. Embedded replicas take this further: the database file lives inside your application process, so reads happen at memory speed with no network hop at all, and changes sync in the background. The second idea is database-per-tenant. Because a SQLite database is just a file, creating one is cheap in a way that spinning up a Postgres instance is not. Teams building multi-tenant SaaS use Turso to give every customer their own isolated database rather than adding a tenant_id column to shared tables - which makes data isolation, per-customer backups, and deletion requests structurally simple instead of a query-level discipline you have to enforce forever. Pricing is freemium with an unusually generous free tier, which is why Turso shows up so often in side projects, hackathon builds, and edge-deployed apps on platforms like Vercel, Cloudflare Workers, and Fly. Paid plans scale on storage, row reads and writes, and the number of databases and locations. The honest limits are the limits of SQLite itself. Write throughput is bounded by a single primary, so write-heavy workloads with high contention are a poor fit, and you give up the extension ecosystem, rich types, and mature tooling that come with Postgres. Compared with Neon or Supabase, which offer serverless Postgres with branching and a broader feature surface, Turso trades capability for read latency and per-tenant economics. Compared with running SQLite yourself on a single server, it trades simplicity for durability, replication, and the ability to survive that server. Choose it when reads dominate, when your users are geographically spread, or when you want one database per customer without one bill per customer.