Muse Image is the first image generation model built in-house by Meta Superintelligence Labs, the AI division led by Alexandr Wang. It creates, edits and combines pictures from plain-language prompts, right inside the Meta AI app, Instagram Stories and WhatsApp. Everyday image making is free, with a Meta One subscription kicking in past an unpublished cap.
Muse Image works as an agent rather than a straight text-to-pixels pipeline. Before drawing, it teams up with the Muse Spark language model to break down your prompt, run a web search when facts matter and plan the composition.
It also fixes its own drafts. A small flaw triggers a local edit; a bigger miss sends it back to a fresh attempt. Meta says nobody coded this behavior, it emerged during reinforcement learning. One happy side effect: lettering comes out spelled correctly, and generated QR codes actually scan.
On Arena's human-preference rankings, the model sits at No. 2 for text-to-image, just behind GPT Image 2.
Access runs entirely through Meta's own apps, with nothing extra to install.
Everyday creation costs nothing. Meta has never published the exact free cap: once you hit it, you either wait for the reset or subscribe to Meta One. Brands get their own version through the Advantage+ ad tools.
| Platform | Status |
|---|---|
| Meta AI app and meta.ai | Live |
| Instagram Stories | Live in the US |
| Live in select countries | |
| Facebook and Messenger | Announced, on the way |
| Advertisers (Advantage+) | Rolling out |
Every picture carries Content Seal, an invisible watermark that survives cropping, compression and even screenshots. Meta plans to extend it to video and is previewing a public detection tool.
One launch feature caused a storm. The option to @-mention a public Instagram account and generate images from its photos was pulled within days after a backlash over consent. It is gone now, though a quick pass through your privacy settings remains a smart habit if your account is public.
Yes, everyday image creation costs nothing in the Meta AI app, on Instagram and in WhatsApp. Meta has not published a precise quota, so you generate freely until a variable cap. After that, two paths open up: wait for the reset or move to a Meta One subscription for heavier use.
Open a chat with Meta AI inside WhatsApp and describe the picture you have in mind in plain language. The result lands straight in the conversation, ready to download or share. Availability is limited to select countries for the moment, while the Meta AI app and meta.ai cover a wider footprint.
No, not anymore. The feature letting anyone @-mention a public account and generate images from its photos was removed days after launch, following criticism over consent. Meta admitted the feature missed the mark. Reviewing your privacy settings is still worth a few minutes if your account is public.
Muse Spark is the large language model from Meta Superintelligence Labs, while Muse Image handles picture generation. The two work as a pair: Spark reasons through your request and plans the scene, then Muse Image renders the final result. A third model, Muse Video, is in preview for video with native audio.
Verdict: Muse Image suits anyone who wants solid AI images without installing a thing, straight from WhatsApp or Instagram. Developers and studios after an API should check back later.
