Muse Code is Meta's terminal coding agent, built by Meta Superintelligence Labs. It takes on complete software engineering tasks across large repositories: planning changes, writing code, and validating the results with persistent background agents. Terminal only, macOS and Linux (Windows via WSL2).
It runs on Meta's Muse models. The Wonder series, Muse Wonder and Muse Wonder Full, is available through Muse Code and the Meta Model API. Meta built these models with two ideas that matter for agent work: Active Memory, the ability to carry what matters across conversations and bring back the details relevant to the task, and Active Focus, the ability to keep working toward your goal when a task gets hard, try other approaches, recover from errors, and check in before following through on anything important.
Pricing is the aggressive part. Monthly plans run Everyday at $5, High at $15, and Power at $50. Or you can pay per token: Standard at $1.25 per million input tokens and $4.25 per million output, or Contributor at $0.10 and $0.20, with the catch that Contributor sessions may be used to improve Muse Code. No monthly subscription is required if you go the token route.
Five dollars of Muse Code is just enough to give you a taste. Yesterday I burned the whole window on an over-ambitious first prompt. Today they reset my usage and I ran it right: one bug, one session. It matched my coding standards and the fix is deployment-worthy.
Consider me teased. Upgrading to the $15 tier to see what the real thing feels like.
