xAI puts grok-build-0.1 on the API and opens Grok Build to all SuperGrok and X Premium+ subscribers
TL;DR
On May 29, 2026, xAI made grok-build-0.1 — the dedicated coding model behind its Grok Build CLI — available on the xAI API in public beta, following the May 25 broadening of Grok Build itself to all SuperGrok and X Premium+ subscribers. grok-build-0.1 is a fast agentic coding model with a 256K context window, priced at $1 per million input tokens and $2 per million output (with a $0.20 cached-input rate). Grok Build is xAI's entry into the terminal coding-agent category that Claude Code and Codex already occupy.
May 29, 2026
grok-build-0.1 released on the xAI API in public beta — the dedicated coding model behind the Grok Build CLI
$1 / $2
grok-build-0.1 pricing per million input / output tokens, with a $0.20 per million cached-input rate, per xAI
256K
Context window of grok-build-0.1, per xAI
SuperGrok / X Premium+
Grok Build CLI access broadened to all SuperGrok and X Premium+ subscribers, beyond the initial top-tier-only beta
On May 29, 2026, xAI put grok-build-0.1 — the dedicated coding model behind its Grok Build CLI — on the xAI API in public beta, following the May 25 broadening of access to the Grok Build terminal coding agent itself to all SuperGrok and X Premium+ subscribers (the agent launched in early beta on May 25, after an earlier top-tier-only release). Grok Build is xAI's direct entry into the agentic-coding-CLI category that Anthropic's Claude Code and OpenAI's Codex already occupy: a command-line agent that plans, edits files, and runs commands from natural-language prompts, now backed by a model developers can also call directly.
The model: grok-build-0.1. According to xAI, grok-build-0.1 is a model built specifically for fast, agentic coding — covering web development, debugging, and tool use. It has a 256K-token context window, accepts text and image input, and is priced at $1 per million input tokens and $2 per million output tokens, with a discounted $0.20 per million rate for cached input. xAI describes it as served at over 100 tokens per second. That positions it as a low-cost coding model — materially cheaper per token than frontier models like Claude Opus, reflecting its "fast" rather than "frontier" design point. The model is the engine behind the Grok Build CLI and, as of the May 29 API release, is also available directly through the xAI API (in public beta) for developers building their own agentic coding pipelines.
What Grok Build is. Grok Build is a terminal-native coding agent, not an IDE or a chat product. xAI describes it as bringing the agentic experience to your terminal: you describe a task in natural language, and Grok Build produces an implementation plan, which you can approve, comment on step by step, or rewrite entirely before execution begins. It then edits files and runs commands to carry the plan out. This is the same interaction model that Claude Code and Codex use — the agent works in your actual project directory rather than in a sandboxed chat window. xAI says existing tooling — AGENTS.md files, plugins, hooks, skills, and MCP servers — all work out of the box, which lowers the switching cost for developers already running Model Context Protocol servers with another agent.
Parallel execution. For larger tasks, xAI says Grok Build delegates work to specialized subagents that run in parallel, with worktree support so subagents can operate in their own working copies. xAI does not publish a fixed concurrency number in the announcement, and the exact limits are still moving in this early-beta phase. This is the same architectural direction now appearing across the category: Anthropic's Dynamic Workflows fans work across many subagents, and Google's Antigravity CLI spins up specialized subagents under terminal sandboxing. The shared trajectory is clear — the frontier coding tools are converging on parallel subagent orchestration as the way to handle codebase-scale tasks rather than single-threaded prompt chains. (xAI's own documentation is the authoritative reference for current concurrency and worktree behavior.)
Subscription access. Grok Build is gated behind xAI's subscription tiers. Following the broadened rollout, it is available to SuperGrok and X Premium+ subscribers — not only the higher-priced SuperGrok Heavy tier that the initial beta required. The grok-build-0.1 model is separately available to developers through the xAI API, independent of the CLI, for anyone who wants to call it directly rather than drive it through the terminal agent.
Where it sits in the landscape. Grok Build enters a crowded and fast-moving category. Claude Code, OpenAI Codex, Cursor's CLI, and Google's Antigravity CLI all ship terminal agents with overlapping feature sets — plan-first execution, parallel subagents, project-directory editing. Grok Build's clearest differentiator today is price: at $1/$2 per million tokens for grok-build-0.1, the per-token cost of running the agent is well below frontier-model pricing, which matters for high-volume agentic workloads where token consumption scales with task size. Its clearest limitation is maturity — it is an early beta, and xAI's daily-release-notes cadence underscores that the API surface, behavior, and feature set are still moving.
Who it affects. For developers already paying for SuperGrok or X Premium+, Grok Build is now included — worth a look if you run agentic coding tasks and want a cheaper per-token option for high-volume work. For developers building their own pipelines, grok-build-0.1 on the API is a low-cost coding model worth benchmarking. For teams standardized on Claude Code or Codex, the practical question is whether grok-build-0.1's lower token cost offsets the maturity gap against more established tools. See our Claude Code review for the incumbent agentic-CLI baseline, our Cursor review for the IDE-native alternative, our Cursor vs Claude Code comparison for the head-to-head on agentic coding, and our best AI coding tools roundup for the full landscape.
Why It Matters
The agentic coding CLI is now a four-way race, and xAI is competing on price. Grok Build does not introduce a new interaction model — plan-first execution, parallel subagents, MCP support, and project-directory editing are now table stakes shared with Claude Code, Codex, and Antigravity. What it changes is the cost floor: grok-build-0.1 runs at $1/$2 per million tokens, well below frontier-model pricing, which matters specifically for high-volume agentic work where token consumption scales with task size. The May 29 API release puts that model in developers' hands directly — alongside the May 25 broadening of CLI access beyond the top-priced tier — lowering the barrier to try it both ways. The honest caveat is maturity: Grok Build is an early beta, so its behavior and API surface are still moving. The signal worth noting is the convergence — every major lab now ships a terminal coding agent built on parallel subagent orchestration, and the competition is shifting from whether the architecture works to price, reliability, and ecosystem fit.
Who's Affected
- — Developers already subscribed to SuperGrok or X Premium+. Grok Build is now included with your subscription. If you run agentic coding tasks, it is worth testing — particularly for high-volume work where grok-build-0.1's $1/$2 per million token pricing is cheaper than running frontier models. Treat it as an early beta: start with non-critical tasks.
- — API developers who want a cheap coding model. grok-build-0.1 is available directly via the xAI API at $1/$2 per million tokens (256K context, $0.20 cached input), independent of the CLI. If you are building your own agentic coding pipeline and want a low-cost model, it is worth benchmarking against other budget coding models.
- — Teams standardized on Claude Code or OpenAI Codex. The practical evaluation question is whether grok-build-0.1's lower per-token cost offsets the maturity gap. Grok Build is newer and its behavior is still stabilizing, so do not migrate critical workflows on the basis of price alone yet.
- — Developers tracking the agentic-coding-tool race. Grok Build is the fourth major lab (after Anthropic, OpenAI, and Google) to ship a terminal coding agent built on parallel subagent orchestration. If you are evaluating where the category is heading, xAI's entry — competing primarily on token price — is a useful data point on how the tools are starting to differentiate.
What To Do Now
- 1. If you already pay for SuperGrok or X Premium+, install Grok Build and run a small task before forming an opinion. It is now included with your subscription, so the only investment is time. Use a non-critical repository for the first run while the early-beta behavior is still settling.
- 2. Lean on the per-token economics, not the feature list, when deciding whether to adopt. Grok Build's feature set largely matches Claude Code and Codex. Its real differentiator is grok-build-0.1's $1/$2 per million token pricing. Model your actual agentic token consumption to see whether the cost difference is meaningful for your workload.
- 3. Reuse your existing MCP servers and agent config. xAI says AGENTS.md files, plugins, hooks, skills, and MCP servers all work out of the box in Grok Build, so tooling you already run with another agent should connect without rework — which lowers the cost of evaluating it in parallel. Check xAI's docs for current concurrency and worktree limits before relying on parallel execution, as those are still moving in early beta.
- 4. Treat the daily-release-notes cadence as a signal to pin versions for anything important. An early beta shipping daily updates means behavior can change between sessions. For any reproducible or production-adjacent workflow, note the version you tested and watch the release notes before relying on consistent behavior.
More on this topic — Best AI Coding Tools
Independent Review
Claude Code
Pricing, pros and cons, real-world verdict — no affiliate spin.
Read the Claude Code reviewMore from ToolNav News