Muse Image (and Muse Video preview)
Meta's first in-house image model behaves as an agent (code, web search, self-critique); Muse Video shown as early preview.
Ends reliance on licensed outside image models inside Meta apps. Muse Image reflects on and edits its own output, supports multi-reference composition, and ranks #2 on Arena image tasks per Meta; Muse Video ranks #3 text-to-video but has audio-sync and fast-motion gaps.
- Date
- Tuesday, 7 July 2026
- Lab
- Meta
- Kind
- model
- Access
- app only
Figures
| Measure | Value | Measured by |
|---|---|---|
| Arena rank (as of 2026-07-05) | Muse Image #2 (text-to-image, edits); Muse Video #3 (text-to-video) Meta-reported leaderboard snapshot | third-party |
TechCrunch reports privacy pushback over tagging public Instagram profiles to generate images of them (default-on, opt-out). Muse Video was not generally available as of the sources opened.
Sources
- ai.meta.com/blog/introducing-muse-image-muse-video-msl/
- techcrunch.com/2026/07/07/meta-rolls-out-muse-a-new-ai-image-generator/
This record was checked against its sources on 6 October 2026. How we check