Zig, Mojo and Gleam: New Programming Languages to Watch

Every year a few new programming languages become fashionable among developers. Your technical team, or a vendor, may mention one of them with real enthusiasm. This post looks at three that are often named: Zig, Mojo and Gleam. For each, we say what it is for, how mature it is as of November 2026, and whether it belongs anywhere near a typical business application. The short version is that all three are interesting, none is a risky mistake to explore, and none is the right default for your next project.

Zig: a careful replacement for C

Zig is a systems language, meaning one for software that sits close to the hardware: databases, operating-system parts, embedded devices and tools. Its own website describes it as having no hidden control flow and no hidden memory allocations, so a reader can see exactly what the program does. It can also use existing C libraries directly, which makes it easier to adopt gradually.

Maturity. Zig is still numbered below 1.0. Version 0.16.0 was released on 14 April 2026, and 0.17.0 followed in early October 2026. We found no statement that a 1.0 release date has been set. A version number below 1.0 tells you the maker still reserves the right to change the language, so code written today may need rewriting later. Its news page also records that the project moved from GitHub to Codeberg in November 2025.

Who uses it. The Zig site lists projects such as the TigerBeetle database and a desktop app called river as examples of what can be built. The Ghostty terminal application also states that its core is written in Zig. These are specialist tools built by enthusiastic experts, not typical business software.

Mojo: Python-flavoured speed for AI work

Mojo comes from Modular, a company building AI infrastructure. According to its FAQ, it began as a way to solve AI programming challenges and is aimed at becoming a general-purpose language. Today it is used for AI algorithms, GPU kernels and other heavy computation, the code that runs on graphics chips to train and serve AI models.

Maturity. Modular announced Mojo 1.0 on 11 August 2026, and its documentation now covers version 1.2.0. The announcement is careful: it says 1.0 does not mean the language is finished, and that changes during 1.x should mainly be additive. Under the stated policy, the core language follows semantic versioning, which is a promise about not breaking existing code without notice. The standard library is a separate matter: its programming interfaces are treated as unstable unless the documentation marks them stable. The language is open source under the Apache 2.0 licence with LLVM exceptions.

The honest takeaway. Mojo is aimed at people writing AI infrastructure. If you run a school or a shop, there is almost no scenario where you would ask a developer to write your billing software in it. If you use AI features, you will use them through existing services; our guide on AI for small businesses in India explains how.

Gleam: a friendly language with strong safety rules

Gleam is a language for building reliable servers. It runs on the Erlang virtual machine, the same technology behind systems built for massive numbers of simultaneous connections, and it can also compile to JavaScript. Its website highlights that it has no null values and no exceptions, two common sources of surprise crashes in other languages, and a built-in toolset for formatting and packages.

Maturity. Gleam version 1 was announced on 4 March 2024, and version 1.19.0 was published on 5 October 2026. It is therefore past the 1.0 mark, which is a stronger position than Zig. Its biggest practical limit is the size of the community: far fewer developers, tutorials and ready-made packages exist than for TypeScript or Python. The sponsors listed on its site include Lambda Class, Atuin, NineFX and n8n, but the pages we read did not name companies running Gleam in production.

What these three have in common

• They are real, actively developed projects with working releases, not fads that disappeared.
• Each solves a narrow problem very well, but none was designed for the general run of websites and management software.
• Each has a small pool of developers, which makes hiring and handover harder.
• Each is changing faster than long-established languages.

Cost and risk: the real conversation

For a business, a language is not a purchase of its own. It is a decision about who can build, fix and extend your system for the next five to ten years. A rare language narrows that group sharply. If your only developer who knows it leaves, the cost of finding a replacement, or of rewriting, can exceed whatever the new language saved. This sits close to what we describe in our piece on technical debt.

The same logic explains why we default to widely used tools. Our own stack is built from Next.js, TypeScript, Node.js, PostgreSQL and Kotlin because every one of them has a deep supply of developers in India and abroad.

A checklist if someone proposes a new language

1. Ask for the specific problem it solves that a mainstream language cannot.
2. Check the version number and the maker's own statement about stability.
3. Ask who else, besides the proposer, could maintain the code in two years.
4. Request a small pilot, not a full commitment.
5. Confirm you get the full source code and documentation at handover.

Where this leaves you

Treat new languages as something to follow, not something to buy. If your project truly needs an unusual tool, we will say so plainly at VS Tech Builders; most of the time the better path is the one in our six-step process, built on proven technology. To compare how we price and plan, see our pricing page.

Sources

Checked on 3 November 2026. Technology changes quickly; always confirm details with the official source before you act on them.

• Ghostty — Ghostty project, GitHub
• Zig News — Zig Programming Language
• Overview — Zig Programming Language
• Mojo 1.0 is here — Modular Forum
• Gleam news — Gleam

Frequently asked questions

Is Zig ready for production use?

Zig is used by some specialist projects, but it is still numbered below 1.0 (0.17.0 in October 2026), so the language can still change in ways that break existing code. It suits expert teams building low-level tools. It is not a sensible default for ordinary business websites or management software.

Is Mojo a replacement for Python?

Not for everyday business software. Mojo is aimed at AI and high-performance computing such as GPU code. Modular announced version 1.0 on 11 August 2026 and describes it as a growing general-purpose language, but Python remains the common choice for scripts, data work and web back ends.

Is Gleam stable?

Gleam passed version 1 in March 2024 and was at version 1.19.0 in October 2026, so the language is past its early phase. The main limits are the small community and fewer ready-made libraries. That makes hiring and long-term maintenance harder than with mainstream languages.

Should my company choose a new programming language to stay modern?

Rarely. A language matters because of who can maintain your system over many years. Mainstream languages have many more developers, libraries and documentation, which lowers risk and cost. Choose a new one only for a specific, measurable need and begin with a small pilot.