A.I.
Meta · Muse
Muse Image
Muse Image is Meta Superintelligence Labs’ image generation model. Meta describes agentic tool use (search, coding tools, self-refine) and integration with Muse Spark. It is available on Meta AI / meta.ai and related surfaces.
- Status
- Current
- Released
- 7 Jul 2026
- Last checked
- 12 Oct 2026newest source date on this page
What it does
Meta launched Muse Image on 7 July 2026 as the first media model from Meta Superintelligence Labs, Meta's frontier AI group (Meta, 7 Jul 2026). It makes images from text, edits images, and combines several reference images, such as a person, a product, and a style, into one picture.
Meta says it works as an agent, a model that takes steps on its own: it can search the web for facts and references, write and run code for accurate charts and QR codes, and check and redo its own drafts. It also plans together with Muse Spark, Meta's chat model. Meta says quality rises the more compute it spends thinking, called test-time compute.
Where it is available
Per Meta on 7 July 2026: the Meta AI app and meta.ai, Instagram Stories in the US, and WhatsApp in some countries, with Facebook coming soon. Meta has not published API pricing for Muse Image in this post.
Benchmarking
Meta says Muse Image held the No. 2 spot on Arena, a public site where people vote between two models' outputs, for text-to-image, single-image editing, and multi-image editing, as of 5 July 2026. That is Meta's report of a human-preference ranking, not a fixed test score.
Watermark
Images made in the Meta AI app and on meta.ai carry Content Seal, an invisible watermark that Meta says survives cropping, compression, resizing, and screenshots. Meta is previewing a tool to check for it. See AI regulation for the EU labelling rules.