xAI launches grok-build-0.1 coding model in public beta at $1/$2 per million tokens

xAI's grok-build-0.1 is a purpose-built coding model now available in public beta through the xAI API, the same model that powers the Grok Build CLI. It's optimized for agentic coding tasks — web development, debugging and MCP support — and emphasizes speed at over 100 tokens per second. The pricing is aggressively low at $1 per million input tokens and $2 per million output tokens.

The positioning is explicitly cost-and-speed rather than peak capability: xAI is targeting general-purpose agentic and tool-calling workloads where throughput and price matter more than topping leaderboards. The CLI-plus-API packaging mirrors the developer-first distribution strategy of Claude Code and OpenAI Codex.

The launch lands amid the week's dominant cost-rationing theme — with WSJ reporting enterprises rationing AI and developers publishing bill-cutting guides, a cheap, fast coding model is well-timed. The competitive question is quality: with the broader Grok line trailing on SWE-bench, grok-build-0.1's appeal hinges on whether it's 'good enough' for routine agentic loops at a fraction of frontier-model cost. That's the same task-dependent model-selection calculus developers are debating around Claude Code's Sonnet-vs-Opus tradeoffs this week.