The Bun team just announced they’re moving from Zig to Rust. And in just six days. six days, the new Rust version passed 99.8% of Bun’s existing test suite on Linux. Let that sink in. Almost 960,000 lines of code. And the tests? Green.
The Big Picture
- That 99.8% test pass in six days? Real. Covering 960k lines of code.
- Why do it? Safety, easier contributor onboarding, and it’s just easier to maintain long-term.
- Oh, and Bun’s owned by Anthropic now. They picked it up in December 2025.
- AI agents, Claude-powered ones, did a ton of the heavy porting work.
- But this is still experimental. No official commitment to finish the migration yet.
- We don’t have real performance numbers yet. That’s the next big thing to watch.
So, What Even Is Bun?
Think of Node.js, but on steroids. Seriously. It’s a runtime, package manager, test runner, and bundler all jammed into one fast, developer-friendly tool. It was created by Jarred Sumner and got famous for being ridiculously quick.
When it launched, people were impressed it used Zig. Zig gave Bun tight, low-level control. You know, C-like speed without actually writing C. But engineering’s a fickle game. Priorities shift.
Zig to Rust. Why the Switch?
Building a runtime isn’t just about raw speed. You’ve got to think about keeping it running for years, getting help from others, and not having to hand-hold every new contributor through Zig’s learning curve.
Jarred Sumner laid out the reasons for eyeballing Rust:
- Memory safety. Rust’s borrow checker catches bugs before they even compile.
- Ecosystem. There are way more crates, better tools, and established patterns.
- More people know Rust: Zig is niche. Rust? More developers have touched it.
- AI works better with Rust: Turns out, Claude “gets” Rust code better than Zig.
They wrote up this massive 300-rule guide for porting Zig to Rust. Then they basically let Claude-powered AI agents loose on the code. Some folks are calling it a “vibe port.”
Six Days. Seriously.
That’s all it took. Jarred tweeted the results and the programming world kind of lost its mind. Reddit blew up. Hacker News threads went wild. 99.8% compatibility in less than a week.
But here’s what most people glossed over: this doesn’t mean Bun is definitely going Rust. Jarred said on Hacker News:
“We haven’t committed to rewriting. There's a very high chance all this code gets thrown out completely. I'm curious to see what a working version looks like, how it performs, and how it feels.”
Classic engineering move. Test the idea, see what happens. Then decide.
How They’re Doing It
Two phases. Simple, right?
Phase A: Just Get It Over
Translate the logic file by file. Don’t even worry if it compiles yet. The point is to capture the intent. AI agents are doing the bulk of this, following that giant rulebook.
Phase B: Make It Actually Run
Get everything compiling, one crate at a time. Run it. Benchmark it. See if Rust can keep up with Zig’s raw speed.
We’re deep in Phase B now. And it’s looking promising.
The Performance Puzzle
Zig still has some aces up its sleeve. Its comptime features and direct hardware control give it a real edge for low-level stuff—memory layout, startup time, raw throughput. Zig has historically been a beast here.
But Rust isn’t just hanging around.
It brings:
- Zero-cost abstractions
- Fearless concurrency
- A huge ecosystem of optimized crates
- Memory safety without a garbage collector slowing things down
The real question: do Rust’s perks outweigh any speed hit? For a project like Bun, where stability and maintainability are just as important as speed… maybe they do.
The Anthropic Angle
Easy to miss: Anthropic acquired Bun in December 2025. That changes things.
Now Bun’s JavaScript runtime, Claude Code CLI, and Anthropic’s LLM services are all under one roof. If you’re running AI workloads with Bun, that’s worth considering. The Rust rewrite doesn’t flip the ownership script, but it does make Bun easier for more people to work on.
And get this—Claude’s grasp of Rust was key to pulling off the port. We might be watching the early days of AI-assisted language migrations becoming a real thing.
What Does This Mean for You?
If you’re using Bun today, chill. Your code works.So packages install.Yet tests run. Nothing breaks.
But looking down the road:
- Stability could get a boost. Rust’s safety nets mean fewer weird runtime crashes.
- More hands on deck. Lower barrier for new contributors.
- Longer life ahead: A bigger ecosystem of tools and libraries.
- Embedding possibilities: They’re actively looking at letting you embed Bun inside Rust apps natively. That could open up wild new uses—think test automation, scripting layers, plugin systems.
Questions You Might Have
Is Zig losing to Rust?
Nah. Different tools for different jobs. Bun’s looking at switching because of what they need—not a universal verdict.
Will the Rust version be slower?
Too soon to tell. Phase B benchmarks will spill the real tea. But Rust holds its own against C and Zig in most cases.
Should I jump from Node.js to Bun now?
If speed’s your thing and you’re okay with a younger ecosystem, give Bun a spin. The Rust rewrite actually makes Bun’s future look more stable, not less.
Why This Experiment Matters
This is bigger than Bun. It’s a live test case for huge codebase migrations, AI-assisted rewrites, and that constant tug-of-war between raw performance and developer sanity.
Whether Bun ends up shipping in Rust, Zig, or some mix, what they learn here will shape how we think about building systems software for years.
I’m waiting on those Phase B performance benchmarks. When they land, we’ll finally have real data, not just speculation.
For now, if you’re even a little curious about Bun, go check it out. The current Zig version is production-ready, crazy fast, and worth your time—no matter what happens with Rust.
So what’s your take? Betting on Bun’s Rust future or sticking with Zig? Let’s hear it.
If you want to dig deeper.
- [Bun's rewrite from Zig to Rust passes 99.8% of testsuite - Reddit]
- [The Great Zig-to-Rust Experiment - Rust Bytes]
- [Bun Is Porting from Zig to Rust - DEV Community]
- Bun's Rust rewrite hits 99.Yet% test compatibility - Hacker News
- Bun official site
- Embedding Bun within other native applications - GitHub Issue