← Back to feed

xAI has launched Grok Build, a coding agent and command-line tool for professional software engineering work, targeting the same market Claude Code and OpenAI Codex have dominated through the first half of 2026.

The product is in early beta, initially available only to SuperGrok Heavy subscribers paying $300 per month — the same tier that provides full access to Grok 4.3 at extended context. xAI described Grok Build as “a powerful new coding agent and CLI for professional software engineering and complex coding work.” The company will collect user feedback during beta before expanding access.

Late to the Race

The timing is pointed. Musk admitted under oath in April that xAI distilled OpenAI models to train Grok, and in a separate filing said xAI was “not built right” when it came to software engineering. Two months later, the company is announcing its own coding agent, competing directly against a market that has had 12 months of head start.

Claude Code, which shipped general availability in early 2026, has become the dominant agentic coding tool at enterprise scale. Uber disclosed burning its entire AI budget by April on Claude Code costs of $500–$2,000 per engineer per month. OpenAI Codex followed with its own cloud-based agent. Both have published benchmark results on Terminal-Bench 2.0 and SWE-bench Verified.

Grok Build has launched without publicly disclosed benchmark scores.

What’s Known

• Access: SuperGrok Heavy subscribers only at launch ($300/month)
• Format: Coding agent plus CLI, matching Claude Code and Codex CLI in product form
• Status: Early beta — xAI explicitly plans to revise based on feedback
• Model: Powered by Grok 4.3, which scored 53 on the Artificial Analysis Intelligence Index at $2.50/M output tokens and holds an estimated tau2-bench Telecom score of 98.0%

xAI’s competitive gap is real: Claude Opus 4.7 leads Terminal-Bench 2.0 at 90.2% (vix scaffold), while the best Grok 4.3 result on the same benchmark has not been published. On SWE-bench Pro, Opus 4.7 leads at 95.7% versus Grok 4.3’s unconfirmed position below the frontier tier.

Why It Still Matters

The coding agent market is a winner-take-most distribution with network effects: the agent that processes more repositories gets better at reading repository-specific patterns, its developers build more integrations, and switching cost grows. xAI entering now — even behind — means it can compete on price, on model capability improvements, and on the natural distribution advantage of the Grok ecosystem across X and SpaceX’s own engineering teams.

The $300/month entry point is also the same as the broader SuperGrok Heavy tier, which means Grok Build is not independently priced at launch. Whether xAI price-competes with Anthropic ($20/month Claude Code tier, pay-as-you-go for Opus 4.7) or targets the same premium enterprise segment is not yet clear.

Benchmark results will determine how seriously the market responds. The press release cycle for coding agents is effectively over — the next data point that matters is Terminal-Bench 2.0 or SWE-bench Pro.