xAI Grok Build 0.1: The Beta Coding Model Powering the Grok Build CLI

xAI has officially released Grok Build 0.1, a specialized coding model now available on Vercel AI Gateway. This beta model is purpose-trained for agentic coding workflows and powers the Grok Build CLI experience.

Key Specifications

Designed for Agentic Coding

Grok Build 0.1 is optimized for the loop of modern software development:

• Planning and architecture
• Code generation and refactoring
• Testing and debugging
• Real-time reasoning traces

The model excels in TypeScript, Python, Java, Rust, C++, and Go - making it ideal for the Grok Build CLI's multi-agent workflows.

Using via AI SDK

import { streamText } from 'ai';

const result = streamText({
  model: 'xai/grok-build-0.1',
  prompt: 'Refactor this module to use async/await and add tests.',
});

Via Vercel AI Gateway

import { streamText } from 'ai';

const result = streamText({
  model: 'xai/grok-build-0.1',
  prompt: 'Create a React component for a todo list with TypeScript.',
});

AI Gateway provides unified API access with:

• Usage tracking and cost management
• Automatic retries and failover
• Performance optimizations
• Built-in observability

Why This Matters

Grok Build 0.1 represents xAI's push into specialized coding models. Unlike general-purpose LLMs, this model is fine-tuned specifically for:

• Multi-agent coordination
• Tool use and code execution
• Long-context codebase understanding
• Parallel subagent workflows

The model powers Grok Build CLI's Plan Mode, Arena Mode, and parallel subagents - making it the engine behind xAI's vision of autonomous software engineering.

Get Started

1. Access via Vercel AI Gateway with model name xai/grok-build-0.1
2. Or install Grok Build CLI: curl -fsSL https://x.ai/cli/install.sh | bash
3. Sign in with SuperGrok Heavy subscription (early beta access)

This model is currently in beta. Reasoning effort is not configurable, and there is no non-reasoning mode available yet.