Meta already owns a large language model (Muse Spark) and sells subscriptions to power users. But it still relied on third-party companies for image generation. Now it has its own tool.
The development of Muse Image was spearheaded by Alexandr Wang's team at Meta Superintelligence Labs, a division formed to accelerate Meta's AI ambitions. This internal effort underscores Meta's commitment to owning the entire AI pipeline, from model training to deployment across its social media ecosystem.
The company has spent heavily on AI infrastructure. It is trying to generate new revenue sources. Muse Image is part of that plan.
With Muse Image, Meta can offer its own image creation. That helps advertisers make ad variations quickly. Meta said in a blog post: "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."
Get your free investing masterclass bonus when you join Market Briefs, our free daily newsletter
The move also opens a new revenue stream. Users can use Muse Image for free through Meta's AI app and website, as well as via WhatsApp's direct messaging and Instagram Stories.
Those who are power users or creators need to subscribe to one of Meta's new monthly plans. Subscribing allows these users to generate numerous AI images and unlock special capabilities. When the free quota is exhausted, they have the option to buy a Meta One subscription or simply wait for the limit to refresh.
The company also aims to branch out beyond its primary advertising operations, creating fresh income streams that offset its massive investments in AI infrastructure. Muse Image will drive tools tailored for advertisers to create images, integrated within Meta's Advantage Plus service. This service enables brands to craft ad creative with greater ease and automate various activities, according to Meta. Meta noted that it collaborated with companies and advertisers during the launch of Muse Image.
Additionally, Meta intends to debut Muse Video, an AI video generator, and stated in a technical blog that it "offers competitive performance in prompt adherence, visual fidelity, and temporal consistency."
Competitors like OpenAI (with its GPT Image 2 model) and Alphabet's Google (with Nano Banana) already offer image generation. Internal benchmarks from Meta indicate that Muse Image lags behind OpenAI's GPT Image 2 but outperforms Google's Nano Banana 2 in editing single and multiple images.
Meta's decision to build its own image model aligns with broader industry trends where major tech firms seek to control more of their AI stack. By reducing dependency on third-party providers, Meta not only cuts costs but also gains tighter integration across its vast ecosystem of apps, potentially improving performance and data privacy for users and advertisers alike.
Subscribe to Market Briefs, our free daily newsletter, and claim your bonus investing masterclass