Meta released a new artificial intelligence image-generation model called Muse Image on Tuesday, part of the company’s push to draw creators and advertisers to its platforms.
The tool, previously known internally by the code name Mango, is the second major product from Meta Superintelligence Labs, the division led by Alexandr Wang.
The unit unveiled its first release, the Muse Spark large language model, in April. That model replaced Meta’s earlier Llama family of AI systems, according to reports.
Consumers can use Muse Image for free through the Meta AI app and website, WhatsApp direct messages and Instagram Stories. Heavy users and creators who want to generate large volumes of AI images or unlock additional features will need to sign up for one of the monthly Meta One subscription plans the company introduced in May. Once free users reach their generation limit, they can either pay for a subscription or wait for the limit to reset, Meta said.
The model will also drive image-generation tools built for advertisers through Meta’s Advantage Plus platform, which is designed to help brands produce ad creative and automate parts of their marketing campaigns. Meta said it worked with businesses and advertisers ahead of the launch.
“Muse Image brings native reasoning to the creative process to adjust elements, swap styles, and create variations based on the advertiser’s creative, resulting in high-quality, on-brand ad variations with fewer iterations,” the company said in a blog post aimed at businesses. Meta said advertisers and agencies can expect to see image variants powered by the new model in the coming weeks.
The launch reflects Meta’s broader effort to diversify beyond online advertising, its core business, and find new revenue streams as it continues to pour money into AI infrastructure.
Meta enters the image-generation market after rivals OpenAI and Alphabet, both of which introduced similar tools earlier. Google’s Nano Banana model, released last fall, proved popular with consumers.
Meta published internal benchmark results showing Muse Image performing behind OpenAI’s GPT Image 2 model but ahead of Google’s Nano Banana 2 in tasks such as editing single and multiple images.
The company has relied on outside AI developers, including Midjourney and Black Forest Labs, to power image- and video-generation features in its Meta AI app and site. Meta said it intends to lean on its own technology going forward to reduce that dependence on third parties.
In a related technical blog post, Meta said it also plans to release a video-generation model called Muse Video at a later date, describing it as offering “competitive performance in prompt adherence, visual fidelity, and temporal consistency.”
Muse Image is expected to expand to Facebook and Messenger, along with additional areas of Instagram and WhatsApp, later this year.
