Zerobrew Promises 20x Faster macOS Package Management Using Rust

Orhun Parmaksız, best known as the maintainer of the Ratatui terminal UI library, has announced Zerobrew, an experimental Rust-written client that positions itself as a drop-in replacement for Homebrew on macOS. The headline claim is aggressive: 20x faster performance. The actual benchmark data tells a more nuanced story, but even the conservative numbers are hard to ignore.

The architecture is the real pitch here. Zerobrew is not a thin wrapper around Homebrew's existing tooling. It rebuilds the delivery mechanism entirely around parallel streaming via Tokio and content-addressable storage (CAS) with O(1) lookups, combined with APFS clonefiles for filesystem-level copying. It still pulls from the existing Homebrew ecosystem, so your formulae and taps remain intact, but everything about how packages get fetched and placed on disk is different.

The distinction between cold and warm installs matters a lot when reading the benchmark numbers. Cold installs, where Zerobrew has no cached artifacts, show roughly 2x improvement over Homebrew. Warm installs, where CAS can serve packages without re-fetching, are where the big gains appear. The per-package warm speedups in the published metric table vary considerably: tesseract drops from 18,950 ms in Homebrew to 643 ms warm in Zerobrew, a 29.5x speedup. Libsodium and sqlite each show 18.1x warm improvement. The top-100-package average warm speedup lands at 7.6x, which is still substantial but well below the headline figure.

Ffmpeg is the honest asterisk in the benchmarks. Cold install time actually regresses slightly, going from 3,034 ms in Homebrew to 3,481 ms in Zerobrew cold, because the CAS hashing step carries real CPU overhead on large archives. Warm installs recover cleanly to 688 ms and a 4.4x speedup, but if ffmpeg cold performance matters to your workflow, that's a data point worth keeping in mind before switching.

The "20x faster" framing in the headline is technically accurate only in the upper-bound sense: tesseract warm actually exceeds it at 29.5x, and some packages like python@3.14 come in at 6.9x warm. Calling the project 20x faster as a blanket claim flattens what the benchmark table actually shows. The more precise characterization, drawn from the Dev.to coverage, is "2x faster cold installs and up to 20x faster warm installs," with real-world results depending heavily on which packages you install most frequently.

Worth noting: Zerobrew is explicitly described as experimental. No repository URL, install commands, or detailed benchmark methodology, including hardware specs, macOS version, network conditions, or definition of cold versus warm, have surfaced in the available documentation. The benchmarks are presented as verified, but the verification artifacts are not yet public. For a Rust project built around a trust-sensitive mechanism like CAS, the security and signing model for fetched artifacts is also an open question.

If you're the kind of macOS developer who has watched "Updating Homebrew..." spin for 30 seconds before every install, Zerobrew is at minimum worth tracking. The architectural approach is credible, the warm-cache gains on packages like libsodium and sqlite are striking, and having a Ratatui maintainer behind it suggests someone who understands Rust's async ecosystem deeply. Whether it's ready to replace Homebrew in a production dotfiles setup is a different question, one that needs an open repo and reproducible benchmarks to answer properly.