Provider
Meta through one endpoint
4 Meta models, billed at Meta's own list price with nothing added - alongside every other provider on the same key, with routing and fallback between them.
Where your requests go, and what happens to them
Served from the United States on the endpoint we call.
Read on Meta's own terms, last checked 2026-08-21 - their published policy. Recorded as a published position rather than a commitment we can make on their behalf, which is why the date is here.
The models, and what they cost
Muse Wonder 1.0
meta/muse-wonder-1.0$4 in · $20 out · 1,048,576 ctx
Reasoning · Huge context · Agentic coding
Muse Spark 1.3
meta/muse-spark-1.3$1.25 in · $4.25 out · 1,048,576 ctx
Agentic coding · Huge context · Tool use
Muse Spark 1.2
Off by defaultmeta/muse-spark-1.2$1.25 in · $4.25 out · 1,048,576 ctx
Agentic coding · Huge context · Tool use
Muse Spark 1.1
Off by defaultmeta/muse-spark-1.1$1.25 in · $4.25 out · 1,048,576 ctx
Coding · Huge context
You do not have to choose Meta for everything
That is the point of a gateway: route the cheap work elsewhere, keep Meta for what it is best at, and fall through to another provider when this one has a bad day - on one key, with one bill and one log.