xAI Launches Grok Code Fast 1: Fast, Low-Cost AI Coding Model

xAI has unveiled Grok Code Fast 1, a new agentic AI coding model built on a 3.14 trillion-parameter Mixture-of-Experts (MoE) architecture. Designed for multi-step workflows, Grok Code Fast 1 can autonomously invoke terminals, grep searches, and edit files within IDEs. It delivers 92 tokens per second and supports up to 256,000-token context windows, making it ideal for large codebases. Available free for one week on GitHub Copilot, Cursor, Cline, Kilo Code, Roo Code, opencode, and Windsurf, it targets small to mid-size development teams. At $0.20 per million input tokens and $1.50 per million output tokens, Grok Code Fast 1 undercuts existing offerings like Copilot Pro. Supporting TypeScript, Python, Java, Rust, C++, and Go, Grok Code Fast 1 aims to move AI coding from helper functions to complete task execution, embedding agentic coding into mainstream developer workflows.
Neutral
This announcement focuses on an AI coding model rather than cryptocurrency or blockchain markets. While Grok Code Fast 1 may boost developer productivity and AI adoption, it has no direct link to crypto trading or token values. Past AI model releases (e.g., GitHub Copilot, OpenAI Codex) had negligible impact on market prices. Traders are unlikely to adjust positions based on a coding tool launch, making the broader market response neutral.