Bun 1.2 Builds Postgres, S3 and a Text Lockfile Into the Runtime

Correction (October 9, 2026): This article was originally published as "Bun 2.0 Ships Native SQL, Hot Module Reload, and a Challenge to Node's Dominance." There was no Bun 2.0 release; the features it described shipped in Bun 1.2, and the benchmark figures and install-time table in that version were not from any source we can identify. We have rewritten the article from Bun's official release notes and moved it to a new address. The old link redirects here.

Bun 1.2, released on January 22, 2025, turns several things JavaScript developers usually install as packages into features of the runtime itself: a Postgres client, an S3 client, and a text-based lockfile. It also changes how Bun measures its compatibility with Node.js.

What's new in Bun 1.2?

The headline additions, as described in Bun's release notes:

• Bun.sql, a built-in Postgres client. MySQL support was listed as coming later.
• Bun.s3, a native API for S3-compatible object storage.
• bun.lock, a text lockfile that replaces the binary bun.lockb as the default for new projects. Existing bun.lockb files still work.
• HTML imports, so Bun.serve can bundle and serve a front end from an HTML file.
• A built-in CSS parser and further improvements to the bundler, test runner and SQLite client.

A Postgres client with tagged templates

Bun.sql uses tagged template literals, so values are passed as query parameters instead of being concatenated into the SQL string:

import { sql } from "bun";

const users = await sql`
  SELECT id, name FROM users
  WHERE created_at > ${new Date("2025-01-01")}
  LIMIT 10
`;

Bun's release post claims Bun.sql is "up to 50% faster at reading rows" than the most popular Postgres clients on Node.js. That is Bun's own benchmark; your results will depend on your queries and network.

Why the text lockfile matters

The old binary bun.lockb could not be read in a pull request, which made dependency changes hard to review and merge conflicts hard to resolve. Bun introduced bun.lock as an opt-in in December 2024 and made it the default in 1.2. Bun also says bun install is 30% faster than in Bun 1.1.

Measuring Node.js compatibility with Node's own tests

Instead of tracking compatibility API by API, Bun now runs Node.js's test suite against itself on every change. According to the release notes, Bun passes over 90% of the Node.js tests for a set of core modules, and the post highlights work on node:http2 servers, node:dgram, node:cluster, node:zlib and C++ addons that use V8 APIs. The post also claims Express can serve requests up to three times faster on Bun than on Node.js.

What this means for teams

For new projects, Bun 1.2 removes a handful of dependencies: a Postgres driver, an S3 SDK, and in some cases a separate bundler. For existing Node.js applications, the useful number is the compatibility work: running Node's own test suite gives a clearer picture of what will break than a feature checklist does. The performance figures are Bun's own measurements, so they are worth reproducing on your own workload before you switch.

Sources: Bun 1.2 release notes · Bun: "bun.lock" text lockfile