Grok Build 0.1 (model)

xAI's dedicated agentic coding model — "the same model that powers Grok Build", the terminal coding agent xAI launched in beta on May 14, 2026 and open-sourced in July. Served as grok-build-0.1, it is "specifically trained for agentic coding tasks, including web development, debugging, and MCP support", accepts text and image input, supports function calling, structured outputs and reasoning, and has a 256K-token context window. xAI says it is "served at a blazing-fast 100+ tokens/second" and "performs best in agentic harness like Grok Build, Cursor, Hermes Agent, OpenClaw, Kilo Code, or OpenCode". It appeared on OpenRouter on May 20 and entered public beta on the xAI API on May 29 at $1 / $2 per million input / output tokens ($0.20 cached; doubled above 200K-token prompts). Parameter count, architecture and training data are undisclosed; weights are proprietary.

The xAI docs list grok-code-fast-1, grok-code-fast and grok-code-fast-1-0825 as aliases, so it succeeds and replaces the August 2025 Grok Code Fast 1 (14 on the AA Intelligence Index v4.3). Artificial Analysis scores the June 16, 2026 snapshot ("Grok Build 0.1 0616") at 27 on the Intelligence Index v4.3 (41 on v4.1.1, where it ranked #65 of 187), characterizing it as above-average intelligence and well priced but "notably slow and very verbose" in its measurements (50 tokens/s output, 0.5 s to first token); xAI publishes no benchmark scores for it. Not to be confused with the Grok Build harness entry, which tracks the Apache-2.0 Rust agent itself.