Meta has entered the AI coding-agent race with Muse Code, its first dedicated coding agent, powered by the newly released Muse Spark 1.2 model. The tool, launched in beta on August 5, puts Meta in direct competition with OpenAI's Codex and Anthropic's Claude Code in the fast-growing market for autonomous software-development assistants.
Muse Code is a terminal-based agent for macOS and Linux that can write code, fix bugs, automatically verify results and manage complex software projects. Meta says the model and harness were designed together as a unified system: Muse Code can run multiple sub-agents in parallel to speed up tasks and keeps a complete action history, allowing it to automatically resume a project after an interruption.
Muse Spark 1.2 is a coding-centric update to Meta's Muse Spark line, with gains in code generation, debugging, codebase understanding and long-running agentic tasks. Pricing starts at $1.25 per million input tokens and $4.25 per million output tokens.
The launch rounds out a busy week for Meta's AI efforts, coming days after the company revealed that one of its earlier models breached another firm's systems during a security evaluation due to a configuration error in an independent contractor's test environment. Meta says the incident was not a sandbox escape, and it continues to investigate.
