Meta has spent years asking to see your camera roll for "memories" purposes. Now it wants to remix that camera roll into whatever surreal, product-shot, or freshly-restyled version of reality you can describe in a sentence. Say hello to Muse Image.

Meta's First Homegrown Image Model Moves In

Announced July 7, Muse Image is the first media generation model to come out of Meta Superintelligence Labs, the unit run by Alexandr Wang. It's rolling out inside the Meta AI app, Instagram, and WhatsApp immediately, with Facebook, Messenger, and advertiser tools arriving via Meta's Advantage+ creative suite soon after. Everyday use is free through the Meta AI app and site, WhatsApp DMs, and Instagram Stories.

The model handles the usual generative party tricks — text-to-image, image-to-image editing, instruction-based edits using plain language, sketches, or handwritten annotations — plus photo restoration, style transformations, room restyling, and product photography, all built to work with your actual personal photos rather than generic stock-art prompts.

An Agent, Not Just a Model

The more interesting detail is under the hood: Meta describes Muse Image as operating agentically, calling out to search and coding tools mid-generation, self-refining its own output, and improving results by throwing more test-time compute at harder requests. That's a meaningfully different architecture than "type a prompt, get a picture" — it's closer to giving the model room to double-check its own homework.

It's also Meta's second big swing from Superintelligence Labs, following April's Muse Spark language model that quietly took over as the successor to the Llama family. Reading between the lines, Meta is building its own full-stack AI identity — model, image, and eventually video with the promised Muse Video — instead of leaning on partner models forever. For a company whose entire business is photos and videos of your life, owning the pipeline that generates and edits them isn't a side project, it's the plan.

Your Instagram Stories are about to get a lot more "AI-restored childhood photo" and a lot less "poorly lit brunch."

Source: Meta AI