Bun 1.1.13 Ships Memory Fixes and Jest-Style Test Sharding as Pulumi Adopts It as a First-Class Runtime
Anthropic's first major Bun release since acquiring the runtime tackles long-running-process leaks, adds isolate, parallel and shard test flags, and lands days after Pulumi made Bun a fully supported infrastructure-as-code runtime.
Overview
Bun, the all-in-one JavaScript runtime, bundler, package manager, and test runner acquired by Anthropic in December 2025, released version 1.1.13 on April 21, 2026. The release targets two areas the project has been criticized for — memory management in long-running processes and gaps in its Jest-compatible test runner — and lands three days after Pulumi made Bun a fully supported runtime for its infrastructure-as-code programs.
Together, the two announcements mark an inflection point for the four-year-old runtime: enterprise tooling is beginning to treat Bun as a peer to Node.js, even as a segment of the production community continues to question its stability.
What Bun 1.1.13 Changes
According to The Register, the 1.1.13 release centers on improvements to the test runner and to memory usage. Bun’s test runner is designed to be compatible with the popular Jest framework, and 1.1.13 adds three flags developers have been asking for: isolate, which runs each test in a fresh environment; parallel, which runs concurrent tests across a user-defined number of processes; and shard, which splits tests across multiple test runners using a syntax matching Jest’s.
Compression in Bun now uses zlib-ng, an optimized fork of zlib, which The Register reports yields up to five times faster compression “without any other code changes.” Bun’s creator Jarred Sumner claims the runtime uses roughly 5 percent less memory in 1.1.13 thanks to an upgraded allocator and an implementation of the Libpas scavenger that returns unused memory to the operating system faster. In Sumner’s words, quoted by The Register, the two changes “reduce baseline memory usage and fix a class of hangs and crashes in long-running processes.”
That framing is significant because the memory work responds directly to complaints that have accumulated since Bun’s 1.0 release in September 2023. The Register’s report cites a Reddit user arguing that “Bun is not stable enough for production nor faster than Node in production” and points to OpenCode founder Jay V, who said his team is “moving to Node and Electron, away from Bun and Tauri,” blaming “memory issues, crashes, and terrible Windows support.” By naming long-running-process hangs and crashes in the release notes rather than burying the fix in a changelog, Anthropic is acknowledging those complaints directly.
Pulumi Makes Bun a First-Class Runtime
The release arrived three days after Pulumi announced on April 18 that Bun is now a fully supported runtime for its infrastructure-as-code programs, going beyond the project’s previous role as merely an alternative package manager for Pulumi’s TypeScript projects. According to InfoQ, Pulumi 3.227.0 lets developers set runtime: bun in Pulumi.yaml and have Bun execute their entire infrastructure program — no Node.js installation required.
InfoQ reports that the integration was among the most-requested features in Pulumi’s GitHub issue tracker since Bun’s 1.0 release. The publication highlights three advantages Pulumi cited for Bun users: native TypeScript execution with no ts-node or separate compile step, faster dependency installation that speeds up CI/CD bootstrapping, and compatibility with the npm packages Pulumi users already depend on. InfoQ cites Pulumi’s claim of “4x faster startup times (5–15ms vs 60–120ms) and 6–35x faster package installs compared to Node.js.”
The Pulumi integration carries caveats. InfoQ notes that Pulumi’s callback functions — sometimes called “magic lambdas” — are not supported under the Bun runtime because they rely on function serialization that depends on Node.js’s v8 and inspector modules, which are not fully available in Bun. Dynamic providers are unsupported for the same reason. Bun runtime support requires Bun 1.3 or later and Pulumi 3.227.0 or later, per InfoQ.
Why It Matters
Bun occupies an unusual position in the JavaScript ecosystem. It is built in Zig and uses Apple’s JavaScriptCore engine rather than V8, and it bundles a package manager, bundler, and test runner into a single binary. InfoQ notes that Anthropic, after acquiring the project, committed that Bun “will remain open source and MIT-licensed” and would continue to be developed for general JavaScript and TypeScript use.
The Machine Herald previously reported that Anthropic’s Bun acquisition in December 2025 was one of a cluster of deals in which AI vendors have absorbed developer toolchains their coding agents depend on — OpenAI’s purchase of Astral, the Python tooling company, followed a similar logic. Bun 1.1.13 is the first major Bun release since that acquisition to directly address the stability criticisms that have dogged adoption, and its arrival alongside Pulumi’s runtime support suggests Anthropic’s stewardship is beginning to show in both code and ecosystem positioning.
What We Don’t Know
Neither article provides independent benchmarks verifying the 5 percent memory reduction or the claim that compression is up to five times faster; the numbers come from Sumner and from Pulumi’s own measurements. It is also not yet clear whether 1.1.13’s memory work is sufficient to change the views of developers like Jay V who have already migrated off Bun. Windows support, which Jay V singled out as a blocker, is not explicitly addressed in the release notes summarized by The Register.
Bun’s trajectory under Anthropic will be tested over the next several releases, as the runtime’s user base expands into infrastructure workloads like Pulumi’s — where long-running processes and memory behavior matter more than raw startup speed.