Grok Build 0.1 is now available on Vercel AI Gateway. If you are building agentic coding tools or experimenting with autonomous code generation, this is a new option worth knowing about.
This is a beta model from xAI, trained specifically for agentic coding tasks. It currently powers the Grok Build CLI app. The model is in early access, so expect rough edges and limited availability.
One thing to know upfront: reasoning effort is not configurable, and there is no non-reasoning mode. The model always reasons. You cannot tune it down for speed or cost. That is a constraint worth factoring into your architecture decisions before you integrate.
To use it via the AI SDK, set the model to xai/grok-build-0.1. That is the full identifier you pass through the gateway.
Vercel AI Gateway itself provides a unified API for calling models, along with usage and cost tracking. If you are already routing other model calls through the gateway, adding Grok Build 0.1 is a straightforward addition to your existing setup.
The practical implication: if you are prototyping an agentic coding assistant or building tooling around autonomous code tasks, spin up a quick integration today using the AI SDK and the xai/grok-build-0.1 model identifier. Use the gateway's usage tracking to keep an eye on costs, especially since reasoning is always on and that can affect consumption. Treat this as an early signal of what xAI is building toward for coding agents, and evaluate it against your specific task requirements before committing it to a production path.