xAI launches Grok Build to challenge Anthropic and OpenAI in coding agents

xAI launched Grok Build on 14 May, a terminal-based coding agent that pushes Elon Musk’s AI company into the contest for tools that operate inside a developer’s workflow. The product is in early beta, gated behind a SuperGrok Heavy subscription. For developers already watching the category, it is one more model-tied terminal assistant to measure against what Anthropic and OpenAI ship.

Price is part of what xAI is selling. PCMag Australia reported that beta access requires xAI’s $US300-a-month SuperGrok Heavy plan — a tier aimed at developers and teams already spending at the top end of the AI subscription market.

In its launch post, xAI described Grok Build as “a powerful new coding agent and CLI for professional software engineering and complex coding work”. The tool stays in the command line, inspects a codebase and handles work beyond autocomplete. According to xAI’s documentation, users tackling bigger jobs should start in “plan mode”, where the agent maps out steps before it touches files.

That workflow is now standard across the category. Engadget and CIO Dive both positioned Grok Build as xAI’s entry into a coding-agent race already being shaped by Anthropic and OpenAI.

Unlike chat-based assistants, CLI agents work across a repository, propose changes and slot into the version-control habits developers already use. Model vendors have focused so heavily on the category for a reason: the terminal sits closer to real engineering work than a standalone chatbot tab.

Why it matters for developers

For Australian developers and startup teams, xAI’s launch matters less as a product announcement than as a signal about where the frontier-model market is going. The largest AI companies are trying to own the whole software-engineering workflow — model, interface, planning layer and subscription. A team weighing Grok Build against Anthropic-style terminal agents or OpenAI’s coding stack is comparing code quality and latency. It is also deciding how much of its daily engineering process it wants running inside one vendor’s product and billing system.

Terminal tools get sticky fast. Once a team relies on one for planning, debugging and repository-wide changes, the cost of switching climbs.

The early-beta setup also tells you something about xAI’s positioning. Restricting access to SuperGrok Heavy subscribers means the first users will mostly be developers already at the top end of the AI subscription market, or teams trialling frontier coding tools before settling on one. xAI is not pitching Grok Build as a mass-market assistant. For now it reads as a premium wedge into professional software workflows.

For local teams, that choice goes beyond developer convenience. Startups and enterprise groups evaluating AI coding tools also weigh cost controls, internal policies and how deep they want one external model provider embedded in their review and debugging routines.

The competition between xAI, Anthropic and OpenAI is moving past chatbots. It is now about how software actually gets built. Workflow design, pricing and model lock-in are part of the product — and Grok Build gives Australian developers one more terminal-based option to measure on those terms.