Meta Muse with Empryo
Meta's Model API serves Muse Wonder, Meta's strongest model for long-running coding work, and Muse Spark, the faster and cheaper one. Both have a 1M-token context, tool calling and reasoning. Empryo connects in two ways. A Model API key bills per token. A Muse Code subscription signs in with a device code and bills to your plan, with the 5-hour and weekly usage shown in Empryo.
- How you connect
- API key
- How you pay
- Per token with a Model API key, or your Muse Code plan with /login muse-code
What you get
Example models
- muse-wonder-1.0
- muse-spark-1.3
- muse-spark-1.3-contributor
- muse-spark-1.2
When to use it
Pick Meta Muse when you want Muse Wonder or Muse Spark straight from Meta, either per token with a Model API key or on your Muse Code plan.
Set it up
Use a Model API keyLink to this section
Create a key in the Model API dashboard at dev.meta.ai. Keys start with LLM_. Then:
empryo --set-key meta LLM_...
# or, inside Empryo
/login metaMETA_API_KEY works too. Models show up as meta/muse-wonder-1.0 and meta/muse-spark-1.3.
Use a Muse Code subscriptionLink to this section
/login muse-codeEmpryo shows a code. Approve it at auth.meta.com/oauth/device with your Meta account. Models show up as subscriptions/muse-code/<model>, and /usage shows the 5-hour and weekly windows with their reset times. Business and Enterprise (MMA) accounts cannot use this sign-in: use a Model API key instead.
VerifyLink to this section
empryo --list-providers # meta shows as Ready
empryo --list-models metaKeys or subscriptions
Meta Muse connects the way marked below. You can mix ways across task-router slots.
- API keyMeta MusePaste a key once. Empryo keeps it in your system keychain, and you pay the provider for what you use.
- Sign inApprove Empryo in your browser and use a plan or credits you already have. There is no key to copy.
- On your machineRun open models on your own computer. No key, no per-token bill, and nothing has to leave your machine.
- Any compatible APIOne block of JSON config turns any OpenAI-compatible endpoint into a provider, including your own server.