Vercel AI Gateway Adds Grok Build 0.1 for Agentic Coding.

Vercel added xAI's Grok Build 0.1 to AI Gateway this week, expanding its model roster with a beta-stage model purpose-built for agentic coding workflows.

Grok Build 0.1 is the reasoning model that powers xAI's Grok Build CLI tool, now in early access. Unlike general-purpose models, it is trained explicitly for coding tasks that involve multi-step orchestration, refactoring, and test generation. The model currently runs in reasoning mode only, with no option to toggle reasoning effort or disable it entirely.

Developers using the AI SDK can call Grok Build 0.1 by setting the model parameter to xai/grok-build-0.1. The unified API means switching between this model and others from OpenAI, Anthropic, or Google requires no changes to application logic, just the model identifier.

Vercel built AI Gateway to solve a specific problem: testing and deploying AI features across multiple providers without rewriting code. The platform offers a single endpoint that routes requests, handles failover, and surfaces observability data across hundreds of models.

For teams running agents that write, test, and ship code autonomously, this abstraction layer becomes infrastructure. Vercel reported that over 30% of deployments on its platform are now initiated by coding agents, including Cursor, Claude Code, Lovable, and v0. That number has grown 1000% in six months, according to data shared in Vercel's recent infrastructure analysis.

Grok Build 0.1 joins a roster that already includes GPT-4, Claude 3.5 Sonnet, Gemini 1.5 Pro, and dozens of others. AI Gateway tracks usage and cost per model, configures retries and rate limits, and routes requests to fallback providers if one goes down. Teams on Observability Plus get built-in tracing for AI SDK calls, letting them debug agent behavior in production.

Vercel is positioning AI Gateway as part of a broader shift toward what it calls agentic infrastructure: systems designed not just for humans deploying code, but for machines that write, test, and ship software autonomously. The company noted that projects deployed by agents are 20 times more likely to call AI inference providers than those deployed by developers directly.

The addition of Grok Build 0.1 reflects this trend. While the model is still in beta and lacks some configurability, its inclusion signals that Vercel is tracking demand from teams building on cutting-edge reasoning models. For developers already using xAI's API or testing Grok Build's CLI, the Gateway integration removes friction when evaluating performance or switching providers.

Vercel also updated its CLI this week to pull anomaly alert details with AI investigation results inline, letting agents act on production issues without leaving the terminal. Combined with AI Gateway's bring-your-own-key support, the platform is optimizing for workflows where autonomous systems manage deployment, testing, and incident response end-to-end.

Grok Build 0.1 is available now on AI Gateway for teams using the AI SDK. xAI has not yet announced general availability or pricing details beyond the early access program.