Muse Wonder Full moves to subscribers first
Fourteen months of overhaul precede the rollout, with Muse Wonder Full moving to subscribers first.
| Muse Wonder | Muse Wonder Full | |
|---|---|---|
| Availability | Meta Model API and Muse Code | Muse Code subscribers |
| Training site | Prometheus data center | Prometheus data center |
| Training scale | gigawatt-scale | gigawatt-scale |
| Benchmark result | state-of-the-art | state-of-the-art |
| Release date | October 12, 2026 | October 12, 2026 |
The API carries Muse Wonder. Muse Code carries Muse Wonder Full.
Muse Wonder will soon power the Muse agent, starting with subscribers.
The research blog ties the training to Meta's gigawatt-scale Prometheus data center.
The Verge calls the launch a way to make up for lost time. The desk reads that as a distribution constraint, not a missing model.
The document does not prove the models beat GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1, or Gemini 4 Argon. It gives no exact benchmark score, parameter count, context window, price, rollout date, or training compute.
The release is a distribution step, not a performance table.
After research.meta.ai. We did not report this. The pictures, if any, are theirs.
