Muse Code Pricing
How Muse Code bills — pass-through of Muse Spark 1.2 tokens, per session and per month, with worked examples.
Muse Code pricing is a straight pass-through of Muse Spark 1.2 token pricing. The CLI itself is free to install; you pay only for the model tokens the agent burns.
This page walks through session-cost math for realistic team workflows: PR review, targeted edits, and long-horizon refactors.
Every price below is quoted verbatim from Meta's 2026-08-05 launch materials.
How Billing Works
Muse Code passes through Muse Spark 1.2 token pricing. The CLI has no separate license fee at launch.
Standard tier is $1.25 per million input tokens, $0.15 per million cached input, and $4.25 per million output. Meta does not use Standard data.
Contributor tier is $0.10 per million input, $0.002 per million cached input, and $0.20 per million output. Meta uses Contributor prompts and outputs to improve their products.
- CLI: free to install
- Model: pay per token on Muse Spark 1.2
- Standard: $1.25 / $0.15 / $4.25 per M
- Contributor: $0.10 / $0.002 / $0.20 per M
Trying to predict Muse Code spend before you roll it out? Book a consult and we will build the cost model on your real workload.
Book a ConsultationPR Review Session Math (Standard Tier)
A Muse Code PR-review session that reads a 40k-token diff and writes 6k tokens of review costs $0.076 on Standard: $1.25 * 0.040 + $4.25 * 0.006.
Cache hits change this materially. If 30k of the 40k input tokens are cached (the repo context), the cost drops to about $0.03.
Scale to 200 PR reviews per month per engineer and you land around $6-$15 per engineer per month on Standard — well below a per-seat SaaS coding tool.
- 40k-in / 6k-out PR review: ~$0.076 uncached
- With cache hits: ~$0.03
- 200 PRs/mo/eng: ~$6-$15 per engineer
- Cheaper than most per-seat SaaS coding tools
Long-Horizon Refactor Math
A multi-file refactor session that reads 200k tokens and writes 30k costs about $0.38 on Standard uncached, or roughly $0.02 on Contributor.
Multi-agent runs consume more tokens than a single-agent CLI because workers and reviewers issue parallel requests. Budget 1.5-2x the naive estimate for real usage.
Persistent memory reduces re-read cost between sessions. A session resumed the next day does not re-load context from scratch.
- 200k-in / 30k-out refactor: ~$0.38 Standard
- Same task on Contributor: ~$0.02
- Multi-agent overhead: budget 1.5-2x naive
- Persistent memory cuts re-read cost
Team-Scale Forecast
A five-engineer team on Standard tier, running 200 PR reviews and 400 targeted edits per engineer per month, lands around $150-$400/month total.
The same team on Contributor lands around $10-$30/month — but Contributor is only appropriate for non-sensitive work.
When we run our own trend-pass routine across the layer3 portfolio we forecast API spend before rollout and alert on tier-mismatched traffic. Do the same before flipping any team over.
- 5-eng team, Standard: ~$150-$400/mo total
- Same team on Contributor: ~$10-$30/mo
- Contributor for non-sensitive work only
- Alert on tier-mismatched traffic in production
How to use Muse Code
You do not host Muse Code yourself — you use it through a tool, so "getting started" really means choosing the right one.
The fastest way to put Muse Code to work day to day is inside an AI IDE, and Cursor is the most popular — it supports it directly, so you can be working in minutes. Prefer a different editor? Windsurf, Zed, and GitHub Copilot drive these models too.
Frequently Asked Questions
- Muse Code passes through Muse Spark 1.2 token pricing. Standard: $1.25 in / $0.15 cached / $4.25 out per M. Contributor: $0.10 in / $0.002 cached / $0.20 out per M.
- No — the CLI is free to install. You pay only for the model tokens the agent burns.
- A 40k-in / 6k-out PR review costs about $0.076 on Standard uncached, or about $0.03 with cache hits. On Contributor, well under a cent.
- Roughly $150-$400/mo on Standard for a realistic PR + targeted-edit workload. Roughly $10-$30/mo on Contributor for the same workload.
- Yes — budget 1.5-2x the naive token estimate. Parallel workers and background reviewers issue more requests per unit of work.
- Volume discounts and enterprise pricing are not documented at launch. Contact Meta directly for enterprise terms.
Forecasting Muse Code Spend for Your Team?
We size AI coding-agent spend against real workload data and design the tier-routing policy that keeps it predictable. Book a free 30-minute audit.
Book a Free Audit