All news
ai

Two Zig-Based Bun Forks Emerge: Cruller Trims Bun for Produc

24 Jul 2026

Two Zig-Based Bun Forks Emerge: Cruller Trims Bun for Production, Buz Aims for Full Drop-In Replacement

What happened

  • Cruller is a fork of the last Zig-based Bun release, reduced to run already-built production JavaScript servers and ported to vanilla Zig 0.16.
  • Cruller retains JavaScriptCore, Bun.serve, HTTP/1-3, WebSockets, fetch, streams, Blob, Request/Response, static serving, and the module resolver.
  • Cruller removes the package manager, bundler/transpiler, shell, test runner, most CLI dispatch, N-API, SQL clients, archive support, and other dev-oriented subsystems.
  • Zig semantic checks, release builds, CJS/ESM entrypoints, Node path tests, and an HTTP Bun.serve plus fetch() smoke test currently pass for Cruller.
  • Cruller requires Zig 0.16.0 and currently supports only Linux x64 as a build target.
  • Buz is a fork of Bun based on the last commit before Bun's Rust rewrite, rewritten in Zig instead of the original language.
  • Buz achieves sub-1 second incremental builds.
  • The Buz author states the goal is to be a drop-in replacement for Bun with a saner codebase.
  • The Buz author describes Bun as 'the quintessential AI slop project at this point.'
  • The Buz author will not accept human-coded contributions until the project reaches a sane codebase state, relying instead on LLMs to improve the code.
  • The Buz post notes a similar Zig-based Bun fork project (Cruller) has already been posted on Ziggit.

Numbers

  • Cruller ReleaseFast stripped runtime: 73.0 MiB
  • Official Bun 1.3.14 runtime: 88.5 MiB
  • Cruller size reduction vs. official Bun: about 18%
  • Cruller V8 Crypto pure-JS benchmark median: about 2% higher than official Bun
  • Over 11,000 lines of dead code removed from the Bun codebase in Buz
  • Original Bun codebase: approximately 600,000 lines of code

Where sources differ

  • Cruller and Buz take differing scope philosophies: Cruller strips Bun down to a minimal production-only runtime (removing package manager, bundler, test runner, etc.), while Buz aims to be a full drop-in replacement for Bun, implying broader feature retention — sources do not clarify how these approaches reconcile.

Why founders should care

  • The smaller runtime size of Cruller may indicate potential infrastructure cost savings for teams deploying pre-built JS servers, though this is unverified at scale.
  • The early-stage status of both forks suggests founders should treat them as experimental rather than production-ready alternatives to Bun today.
  • The divergence between Cruller's stripped-down scope and Buz's full-replacement ambition could signal fragmentation in the Zig-Bun ecosystem, which may affect long-term tooling choices.
  • Buz's LLM-driven development model may indicate a shift in how some open-source infra projects are maintained, which founders evaluating dependencies might want to monitor for stability implications.
  • Limited platform support (Linux x64 only for Cruller) suggests founders with multi-platform deployment needs may need to wait for broader support before adoption.

Sources