xAI’s Grok Build 0.1 is now available on Vercel AI Gateway — no separate xAI account required, no markup on top of provider pricing.
The model is a beta agentic coding model that currently powers the Grok Build CLI. A few things to know before you reach for it:
- Context window: 256K tokens
- Pricing: $1 per million input tokens, $2 per million output tokens
- Reasoning: always on — no non-reasoning mode, no configurable effort level
- Status: early access beta
To call it through the AI SDK, set the model identifier to xai/grok-build-0.1. That’s the entire integration surface — the Gateway handles the rest.
Vercel’s AI Gateway wraps the model with its standard layer: unified API, usage and cost tracking, retries, failover, custom reporting, and observability hooks. For teams already routing other models through the Gateway, adding Grok Build 0.1 is a one-line change.
NOTE
Reasoning effort is fixed — you can’t dial it down for faster, cheaper responses the way you can with some other models. Factor that into latency and cost estimates for high-volume agentic loops.
The model is live now in early access. Check the Vercel AI Gateway model page for current availability details.