About this model
Grok Build 0.1 is xAI's first model in a dedicated coding family, trained specifically for agentic software engineering rather than general conversation. Released in May 2026 and offered in public beta through the xAI API, it targets multi-step development: autonomously planning, writing, refactoring, and iterating on code, invoking tools, and supporting the Model Context Protocol for web development and debugging. It produces text output across these long-horizon coding loops.
Because this is the family's debut, there is no same-family predecessor to compare against. Functionally, it diverges from xAI's general-purpose flagships such as Grok 4.3, which carries a larger one-million-token context and always-on reasoning, and Grok 4.20 with its two-million-token window and toggleable reasoning. Grok Build 0.1 trades that breadth for a tighter 256,000-token window optimized to hold a mid-sized codebase in memory during long-horizon coding loops.
The model powers xAI's terminal-based Grok Build CLI and runs inside agentic harnesses including Cursor, Kilo Code, OpenCode, and Hermes Agent. It sits alongside multi-agent siblings like Grok 4.20 Multi-Agent in xAI's broader lineup.
This About section is AI-generated from public sources (Claude Opus 4.8), with no human editing. It may contain inaccuracies — verify critical details against the sources listed above.
Data sources: Venice API · HuggingFace · Wikipedia — enrichment updated 1d ago