Meta releases Muse Code, command-line AI for large codebases
Meta just dropped Muse Code, a command-line AI tool built to help software engineers handle big codebases more easily.
Made by Meta Superintelligence Labs, it uses multiple sub-agents running in parallel inside isolated worktrees to keep things fast and logs your work locally so you can pick up right where you left off, even after a break.
Muse Code pay-as-you-go and Contributor Tier
Muse Code runs on the Muse Spark 1.2 model and gives you handy commands like /plan for mapping out tasks or /grill for stress-testing that plan before implementation.
Pricing is flexible: pay as you go at $1.25 per million input tokens (and $4.25 per million output), or choose the Contributor Tier for up to 21 times cheaper rates if you're cool with Meta using your coding data to improve its models.
Heads up, though, the Contributor Tier is capped at 60 requests a minute and might not fit companies working with sensitive info.