Why Zig Stays at 0.16 After Ten Years and Bans AI and GitHub

In 2018 Andrew Kelley left his job at OkCupid to live off community donations and create a new programming language called Zig . Over the next eight years the language gained significant attention—JetBrains featured him on a podcast, Stack Overflow ranked Zig as the fourth most admired language, and companies such as Uber, Bun, and Anthropic adopted it—yet its version number remains at 0.16.

Kelley explains that a 1.0 release would lock in design decisions forever, and if those decisions later prove wrong the language would be forced to live with a flawed architecture for decades. He points to the longevity and stability of C as evidence that a language’s competitive advantage can be its unchanging core.

"1.0 is a promise. Once released you lock yourself in. If the locked design is wrong you have to live with it for 50 years."

His commitment to stability leads him to reject modern languages that did not meet his strict criteria: Go introduced garbage‑collection pauses unacceptable for real‑time audio, C++ caused prolonged memory‑corruption debugging, and Rust’s strictness slowed development of a font‑rendering feature. The solution, he decided, was to write his own language.

The philosophy of “slow is better” extends to community management. Zig’s code‑review process is deliberately mentor‑driven: newcomers submit imperfect patches, and core maintainers provide line‑by‑line feedback, turning apprentices into reliable contributors. Kelley argues that AI‑generated contributions are "negative value" because they add no learning for the reviewer and flood the project with low‑quality pull requests.

"AI‑assisted contributions are not zero value. They are negative value."

In March 2026 a survey of 112 major open‑source projects found only four that completely forbid AI contributions; Zig was one of them. The ban is codified in the project’s code‑of‑conduct, prohibiting AI‑generated code, AI‑written comments, and even AI‑suggested design ideas.

In November 2025 Kelley moved the Zig repository from GitHub to Codeberg, a German non‑profit platform, citing three reasons: unreliable GitHub Actions (which he dubbed “vibe‑scheduling”), GitHub’s aggressive promotion of AI tools that conflicted with Zig’s policy, and a fundamental distrust of Microsoft, the owner of GitHub. He also described GitHub Sponsors as a "liability" and preferred to receive fewer donations on a platform he trusted.

"We write software here. If CI doesn’t work, we switch to one that does."

Financially, the Zig Foundation operates on an annual budget of $670,000, with Kelley’s personal salary at $154,000, all sourced from community donations rather than venture capital. He emphasizes that he would rather earn less than be tied to a platform he does not trust.

When asked if he could return to a traditional job, Kelley replied that he "basically can’t be hired" because most companies now require AI‑generated code, which he refuses to use. He also highlighted his preference for open tools: he codes in a terminal with Vim and avoids any closed‑source software, even while speaking on a JetBrains podcast.

Despite the surrounding hype of AI‑driven productivity, Kelley maintains a deliberately slow development cadence: ten years without a 1.0 release, meticulous manual code reviews, and a mentorship model that prioritizes human learning over rapid feature delivery.

He describes Zig as "a temple to the computer" and acknowledges that building such a temple over a decade is a rare endeavor in today’s fast‑moving tech landscape.