Meta rolls out Muse Image AI model that lets you incorporate other Instagram users into generated photos

Meta has unveiled Muse Image, the first AI image-generation model from its Superintelligence Labs division. The tool is now live in Meta AI, Instagram, and WhatsApp, with broader rollout coming to Facebook and Messenger soon.

What sets Muse Image apart#

The model operates with what Alexandr Wang, who leads Meta's Superintelligence Labs, describes as "agentic" capabilities—it works alongside Meta's Muse Spark language model to analyze prompts, search the web, and plan its approach before generating images. This represents part of Meta's shift toward the Muse family of AI models, which are gradually displacing the company's earlier Llama lineup.

A noteworthy feature allows users to @ mention other Instagram accounts directly within their prompts, enabling the AI to weave those people's likenesses into the generated output. Meta draws on public photos to construct these visual representations, though it provides controls so individuals can manage how their content gets reused for AI purposes.

Practical capabilities and rollout#

Beyond incorporating people into images, users can transform existing photos using suggested edits, generate designs for items like invitations and postcards, and even reimagine interior spaces by pulling room photos from Facebook Marketplace or other web sources. The tool also supports direct image editing—users sketch changes onto photos and then share the results to their feed, stories, or messages.

Meta is also preparing a Muse Video model in the pipeline. Early previews suggest it will deliver competitive performance on factors like prompt accuracy, visual quality, and temporal consistency across frames.

The Muse Image model will power 30 fresh AI effects rolling out to Instagram Stories in the US first, before expanding internationally and integrating deeper into Meta's app ecosystem.