Meta's Muse Code for Mac Is Up to 20x Cheaper If You Let Meta Train on Your Data

Meta has introduced Muse Code, a terminal-based AI coding agent available for macOS and Linux. The tool is in beta and runs on the new Muse Spark 1.2 model. It marks the company's first major move into agentic coding, directly competing with OpenAI's Codex and Anthropic's Claude Code.

According to Meta CEO Mark Zuckerberg, Muse Code can handle complete software engineering tasks across large repositories: planning changes, writing code, and validating results through multiple persistent subagents. Installation requires a single command in the Mac terminal, with no dedicated app available.

Muse Code comes with built-in skills: /plan turns a task into an approval-gated plan, /grill stress-tests a plan, and /goal works toward successful completion of an objective. The agent maintains an append-only local event log of every model call, tool run, approval, and edit, allowing it to resume from where it stopped after a crash.

Pricing for Muse Spark 1.2 is set at $1.25 per million input tokens and $4.25 per million output tokens. However, Meta offers a significant discount to users who allow their prompts and completions to be used for training future AI models. With the Contributor tier enabled, prices drop to $0.10 and $0.20 per million tokens—12.5 and 21 times cheaper, respectively.

The Contributor tier is rate-limited to 60 requests per minute and 2.1 million tokens per minute, while the Standard tier allows up to 3,000 requests and 4 million tokens per minute. The performance gap is clear, but for individual developers and small teams, the lower price may outweigh these restrictions.