Meta debuts Muse Image, Superintelligence Labs’ first AI image model

Meta On Tuesday, it launched Muse Image, a new artificial intelligence model for creating images that the company aims to attract creators and advertisers to its offerings.

The AI ​​technology, originally codenamed Mango, represents the second major release from Meta Superintelligence Labs, led by Alexandr Wang, who oversaw the April introduction of the Muse Spark big language model, which replaced the company’s previous Llama model family.

Muse Image will be available to consumers for free via the Meta AI app and site, WhatsApp direct messages, and Instagram Stories. Power users and creators need to sign up for one of Meta’s new monthly subscription plans, which launched in May, to create lots of AI-generated images and access certain features. The company said users can purchase a Meta One subscription if they reach their free limit or wait until their limit is reset.

Muse Image will also power advertiser-specific visual creation tools as part of Meta’s AI-powered Advantage Plus service, which allows brands to more easily develop ad creative for marketing campaigns and automate certain tasks. Meta said it is working with businesses and advertisers as part of the launch of Muse Image.

“Muse Image brings native reasoning to the creative process to adjust elements, replace styles, and create variations based on advertiser creative, resulting in high-quality, on-brand ad variations with fewer iterations,” the company said in a blog post for businesses. “In the coming weeks, advertisers and agencies can expect to see a range of images powered by Muse Image.”

Its new rendering model and efforts to monetize it show how Meta is trying to expand its core business of online advertising and create new revenue streams tied to its big spending on AI-related infrastructure.

OpenAI And Alphabet As Google’s Nano Banana became popular with consumers with its launch last fall, it has gained an advantage over Meta in offering similar rendering models.

The meta also revealed internal benchmarks that show Muse Image follows OpenAI’s latest GPT Image 2 model, but outperforms the Nano Banana 2 model in tasks like editing both single and multiple images.

The social media giant has previously used third-party AI models such as Midjourney and Black Forest Labs to power various image and video rendering features on its Meta AI app and site. The company said it plans to use the new artificial intelligence model to reduce dependence on similar third-party technologies.

Meta also plans to release an AI video rendering model called Muse Video at a later date, adding in a technical blog that it “delivers competitive performance in terms of fast adaptation, visual fidelity, and temporal consistency.”

Muse Image will be available in more areas on Facebook and Messenger, as well as Instagram and WhatsApp services, later in the year.

WRISTWATCH: Three major narrative breaches of AI.