Pricing utility · official current rates

Muse Code Cost Calculator

Estimate current Meta Muse Code costs from input, cached-input, and output token counts across Standard and Contributor pricing tiers.

LIVE ESTIMATEUSD · RATES VERIFIED AUG 09 · 2026

Pricing tierStandardDefault estimateContributorLower rates · policy tradeoff

Presets

Input tokens$1.25 / 1M0 to 1,000,000,000,000 tokensCached input tokens$0.1500 / 1M0 to 1,000,000,000,000 tokensOutput tokens$4.25 / 1M0 to 1,000,000,000,000 tokens

MODELmuse-spark-1.2

Input cost: $0.00
Cached input: $0.00
Output cost: $0.00

ESTIMATED TOTAL$0.00

Meta says this tier is not used to improve its products.

Meta official product pricing source

Estimate only. Token counts, retries, agent activity, and prices can change the final bill.

Formula · three token types

How the estimate is calculated

Each token count is divided by one million, multiplied by its current tier rate, and added to the total. Inputs are clamped to non-negative whole tokens.

(input ÷ 1M × rate)+(cached ÷ 1M × rate)+(output ÷ 1M × rate)

Examples · standard tier

Worked estimates

Workload	Input	Cached	Output	Estimate
Small fix	150K	25K	25K	$0.2975
Typical edit	1M	250K	150K	$1.925
Repository run	10M	4M	2M	$21.60

Use the estimate as a dated budget input

Muse Code is beta, and the final token mix depends on repository size, generated output, retries, validation, and agent activity. Save the tier and Aug 9 rate date with any budget. Re-check Meta’s current page before committing material spend.

The cheaper Contributor result is not a neutral discount: Meta gives it a different product-improvement label. Read the Contributor decision guide and data and privacy questions before using it with proprietary code.

Quick answers · source bounded

Cost calculator FAQ

Which rates does the calculator use?

It uses the Standard and Contributor input, cached-input, and output rates visible on Meta’s current Muse Code page on Aug 9, 2026.

Why are cached-input tokens separate?

Meta publishes a separate cached-input rate for both tiers, so combining them with ordinary input would overstate cost.

Does the estimate include every charge?

It multiplies the token counts you enter by the published rates. Retries, background-agent activity, changing prices, taxes, or unentered usage can change a final bill.

Why does Contributor show a policy notice?

Meta labels Contributor as used to improve its products, while Standard is labeled not used to improve its products. Cost and data choice belong together.