--- title: "xAI Opens Grok Build 0.1 to Developers at $1 per Million Tokens" subtitle: "xAI opens grok-build-0.1 to all developers via public API -- a purpose-built agentic coding model at $1/M tokens with always-on reasoning and a 256K context window" url: "https://alphasignal.ai/news/xai-opens-grok-build-0-1-to-developers-at-1-per-million-tokens" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "xAI" company_url: "https://alphasignal.ai/company/xai" published: "2026-05-29T16:06:07.000Z" updated: "2026-06-16T20:51:06.978Z" type: "News" topics: - "Api" - "Development" subtopics: - "Code Agents" - "Code Generation" upvotes: 5808 read_time: "5 min" source_url: "https://accounts.x.ai/sign-in?redirect=cloud-console&return_to=%2Fteam%2Fdefault%2Fapi-keys" comments: 0 --- # xAI Opens Grok Build 0.1 to Developers at $1 per Million Tokens > xAI opens grok-build-0.1 to all developers via public API -- a purpose-built agentic coding model at $1/M tokens with always-on reasoning and a 256K context window xAI just made its dedicated coding agent model, [**grok-build-0.1**](https://x.ai/news/grok-build-0-1), available to any developer through the public API. Until now, the model was locked behind a SuperGrok or X Premium+ subscription and could only be accessed through the Grok Build CLI. That wall is now down. ## Not a chat model wearing a hard hat Released in public beta, Grok Build 0.1 is a coding model with a specific and deliberate purpose: to function as an autonomous software engineering agent, not just a code-completion helper. While most language models can answer programming questions or fill in a function body, it is trained from the ground up to handle end-to-end development tasks -- planning what needs to be done, executing the work step by step, and debugging failures along the way. The underlying design premise is that coding agents have different failure modes than chat assistants: they need to hold large codebases in context, reason about multi-file changes, and execute sequences of actions without losing track of what they have already done. xAI built Grok Build 0.1 to address those failure modes directly. ## What's under the hood - **256K context window:** The model features a 256K context window with no text output limit, making it well suited for long-horizon coding and automation workflows. - **Always-on reasoning:** Reasoning is always active on Grok Build 0.1. There is no non-reasoning mode and reasoning effort is not configurable, so every response includes structured analysis before the final output. Think of it as the model quietly thinking through the problem before it writes a single line. - **Multimodal input:** The model accepts text and image input, which means it can read diagrams, UI mockups, and error screenshots alongside source code. - **Native tool use and structured output:** The model supports native function calling and reliable structured outputs such as JSON schemas and typed data formats, making it easy to integrate into automation frameworks, orchestration systems, and developer tools. - **Speed:** According to [xAI's official announcement](https://x.ai/news/grok-build-0-1), the model is served at 100+ tokens per second. - **Rate limits:** The model supports up to 1,800 requests per minute and 10 million tokens per minute, which is generous enough for most production agentic workloads. ## The pricing math Early access is now open, with API pricing at $1 per 1 million input tokens and $2 per 1 million output tokens. Cached prompts cost $0.20 per 1 million input tokens. To put that in concrete terms: analyzing a 10,000-line codebase -- roughly 40K input tokens and 10K output tokens -- costs approximately $0.06. For agentic loops that run dozens of iterations, that is a meaningful cost advantage over frontier models priced at 5-10x more. Grok Build 0.1 operates in that highly efficient "Flash" tier sweet spot. Priced at just $1 per million input tokens and $2 per million output tokens, it is built to be economically viable for agentic loops -- the kind of cost-efficiency needed to let agents run wild. One caveat worth noting: if you exceed the 200K context window, costs double. ## Where you can plug it in Developers can now build directly on top of the same model that powers xAI's Grok Build CLI -- without needing a SuperGrok or X Premium+ subscription. Access points include: - Directly via the [xAI API](https://console.x.ai/team/default/api-keys) (grab an API key from the console) - [OpenRouter](https://openrouter.ai/x-ai/grok-build-0.1) and Vercel AI Gateway - IDE integrations: Cursor, Kilo Code - Agent frameworks: Hermes Agent, OpenClaw, OpenCode A quick call looks like this: ``` curl https://api.x.ai/v1/responses \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $XAI_API_KEY" \ -d '{ "model": "grok-build-0.1", "input": [{ "role": "user", "content": "Refactor this module to use async/await and add error handling" }] }' ``` Third-party access is also available for Grok personal subscriptions: SuperGrok or X Premium users can invoke the model via OAuth in Hermes Agent and OpenClaw. ## Where it fits -- and where it doesn't Grok Build 0.1 scored 70.8% on SWE-Bench Verified -- the industry-standard benchmark that measures how well a coding agent can autonomously resolve real GitHub issues. That is a solid number, but context matters: the benchmark gap is the most significant concern. Grok Build trails the top two competitors by 17-18 percentage points on the primary coding benchmark -- not a rounding error, but a meaningful difference in the probability that a given task completes correctly without intervention. The model is well suited for: - Agentic pipelines and orchestration loops where cost-per-call matters - Multi-step web development and debugging workflows - Teams building their own coding agent products on top of an API - Web development, debugging, and MCP (Model Context Protocol) server support - CI/CD automation via headless mode Grok is attractive for lower-cost API usage, large-context reading, and many low-risk attempts. Claude is usually the safer first choice for difficult coding tasks, mature Claude Code workflows, long refactors, and reliability-heavy engineering work. ## The real unlock: a callable primitive The API release is more significant than it first appears. The API launch opens Grok Build 0.1 to teams that want to embed it in their own applications and pipelines. Full Agent Client Protocol (ACP) support means orchestration platforms can call Grok Build as a primitive -- the same way they call Claude Code or Codex CLI. One unusual design choice: Grok Build ships its system prompts in plaintext. You can read exactly what instructions the agent is operating under. This is rare in the coding agent category -- Claude Code, Codex CLI, and Cursor all treat their system prompts as proprietary. It's a small but telling signal about how xAI wants to position this tool: transparency over mystique. Parallel worktrees, Plan Mode, ACP, headless automation -- xAI is betting that multi-agent parallelism and human-in-the-loop planning will outweigh raw benchmark performance over time. Whether that bet pays off depends on how fast the underlying model improves. The model powering Grok Build is currently a beta release -- the benchmark picture may look different in three months. ## Sources - [X post by @xai](https://x.com/xai/status/2060392249402552457) - [Grok Build 0.1 on API | xAI](https://x.ai/news/grok-build-0-1) - [Grok Build 0.1 - API Pricing & Providers | OpenRouter](https://openrouter.ai/x-ai/grok-build-0.1) - [Grok Build 0.1 by xAI on Vercel AI Gateway](https://vercel.com/ai-gateway/models/grok-build-0.1) - [xAI Opens Grok Build 0.1 to Developers via API - DevOps.com](https://devops.com/xai-opens-grok-build-0-1-to-developers-via-api/) - [Grok Build Review: xAI's Agentic Coding CLI -- ChatForest](https://chatforest.com/reviews/xai-grok-build-terminal-coding-agent-review/)