Meta just unveiled Muse Image, the first image-generation model from Meta Superintelligence Labs (MSL). It is live today inside Meta AI, and it brings a level of creative control that puts it in direct competition with OpenAI’s DALL-E, Google’s Imagen, and xAI’s Grok image generation.
Muse Image is the second model from MSL, following Muse Spark in April. Where Muse Spark focused on reasoning and text, Muse Image is built from the ground up for visual creation. And unlike some competitors that require a separate app or website, it is embedded directly into Meta AI, meaning hundreds of millions of users across Instagram, WhatsApp, and Facebook can access it right now.
I covered Meta Compute last week and noted that MSL (led by Daniel Gross) was quickly becoming Meta’s center of gravity for AI. Muse Image confirms that trajectory.
What Muse Image Does
The headline feature is that Muse Image does not just generate images from a prompt. It plans the image first. By pairing with Muse Spark, the model takes multiple behind-the-scenes steps before rendering: planning the layout, looking up real-time web context, and blending multiple visual references into one coherent output.
You can upload existing photos and use them as inputs. The model supports:
- Photo blending — combine multiple photos into one high-quality composite
- Sketch editing — draw annotations directly on an image, and Muse Image renders the edit
- @-mention for Instagram — tag an Instagram account to pull public profile photos into your creation
- Presets — restore old family photos, turn yourself into a claymation character, a 16-bit video game hero, or a Renaissance painting portrait
- Room redesign — snap a photo of your room and Muse Image will refurnish it with real products from the web or Facebook Marketplace
These are not toy examples. The preset list includes professional use cases: product photography, room staging for real estate, architectural visualization, and logo design.
Where It Ships
Muse Image powers more than 30 new AI effects for Instagram Stories, enables image generation in direct chats with Meta AI on WhatsApp (starting in select countries), and is available inside the Meta AI app directly. Facebook and Messenger integration is coming soon.
For advertisers, Muse Image will be available through Meta Advantage+ creative, which keeps it inside Meta’s existing ad ecosystem.
Basic use is free. For users who want more, additional creation capabilities sit behind Meta’s subscription plans.
The MSL Story
Meta Superintelligence Labs was formed after Meta hired Alexandr Wang in a reported $14 billion deal. Muse Spark was their first output in April — a natively multimodal reasoning model. Muse Image is their second, and the company says Muse Video is already in development.
The pace matters. Meta went from zero image-generation model to a fully integrated product in roughly three months. That is what a $14 billion talent investment looks like when it lands.
Caveats
A few things worth noting:
- WhatsApp image generation is limited to certain countries at launch.
- The @-mention Instagram feature requires public profile photos and an opt-in setting.
- Muse Video is announced as “in development” with no release date.
- Meta’s subscription plans for “extra creation” are not fully detailed yet.
Bottom Line
Muse Image is Meta’s most credible move into the AI image generation space. It is not a research demo, a paper, or a blog post. It is live in the most popular social apps on the planet today. The photo blending and sketch editing features are genuinely differentiated from what DALL-E and Imagen offer. The combination with Muse Spark’s reasoning layer gives it a structural advantage in understanding complex prompts.
If you use Meta AI, open it right now and try one of the presets. It works.
