Model board / Muse Spark Open

Muse Spark Open

latest Meta · Llama & Muse

Verified 2026-10-22 against the pages listed below. Prices are USD per 1M tokens, list rate — no committed-use, enterprise or volume discount.

Input
$1.25
Cached input
$0.15
Cache write
—
Output
$4.25
Context window
1,000,000
Max output
—
Weights
open
Caching
unverified

host price (Together AI) — Meta serves no first-party endpoint for the open release; the weights (816B MoE, 42B active) are downloadable. Same rate as closed Spark on Meta's API, so the choice is custody and licence, not price

1M stated by the host; Meta's overview does not restate it. Max output not stated

capabilities as the host describes them; caching behaviour not stated. The licence file ships with the weights and its terms were not re-read this pass

Watch out

Open weights are not open terms: read the licence file beside the weights before building, and price the host rather than assuming Meta's API rate card applies

What we would use it for

Self-hosted or host-routed Meta frontier coding work where the closed API's terms do not fit

What this has cost, week by week

Output price at each refresh since 2026-10-22. A vendor change is the provider changing its price. Our correction means this board changed what it prints and the provider's charge did not move — the two are never merged, because publishing the second as the first would be a price rise that never happened.

10-22 $4.25

Nothing has moved since we started tracking it.

Where these numbers came from

No number on this page was written that was not read from one of those pages on 2026-10-22. If one is wrong, send us the source URL and it is corrected in the next weekly refresh, with a changelog line saying what was wrong — contact@naderu.com.

Discussed in Naderu Weekly

The rest of the board

This row sits alongside 76 others on the model board, refreshed every week. The same data as JSON, and the terms for using it, are at /board-data.