Meta just launched an AI tool that turns text prompts into images, and it costs one cent per image to run. It's called Muse Image, and it comes from Meta Superintelligence Labs, the AI research group Mark Zuckerberg built after paying more than fourteen billion dollars to bring in Alexandr Wang, the founder of data company Scale AI.
There are two versions of this launch. The free one lives inside the Meta AI app, Instagram Stories, and WhatsApp, where anyone can type a prompt and get a picture back. The one that matters more for business is the paid version: a service other companies can plug into their own software to generate images automatically, at scale, for one cent each.
That price is the real story. Building the same kind of picture with rival tools from Google or OpenAI typically costs several cents to over ten cents per image, depending on quality. At one cent, a retailer generating ten thousand catalog photos or an ad agency testing a hundred versions of the same banner suddenly faces a bill in the tens or low hundreds of dollars instead of thousands. That is the kind of cost drop that changes what work gets attempted at all, not just how much it costs.
Muse Image also works differently from a typical image generator. Before drawing anything, it plans out the request, searches the web for real reference material, and in some cases writes and runs small pieces of code to get details like charts or QR codes correct. It checks its own output before handing it back. That extra step is why Meta calls it agentic rather than just a picture generator.
None of this makes Meta the best in the market. By its own published comparisons, Muse Image trails OpenAI's top image model on overall quality. Meta is not trying to win on quality right now. Wang has been explicit that the company's coding tools compete on price rather than capability against Anthropic and OpenAI, and the same pattern is playing out here: cheaper, good enough, and aimed at the huge volume of everyday image work that never needed a premium model in the first place.
There is a catch worth flagging before any company builds on this. The consumer version of Muse Image lets people generate images of real Instagram users by tagging their handle, without notifying the person whose likeness is used, and it has already drawn public complaints over how photos get pulled into the system. Any business planning to use Meta's image tools for customer-facing work, especially anything involving real people's photos, should read the consent terms closely. Meta's privacy record, including a five billion dollar fine from regulators in 2019, means this kind of feature draws scrutiny fast.
For procurement and marketing teams, the practical takeaway is simple: image generation for routine, high-volume work just got a lot cheaper, and that price pressure will likely push other providers to cut prices too.