Meta has launched Muse Image, its first in-house AI image generation model developed by Meta Superintelligence Labs (MSL), embedding it directly into WhatsApp, Instagram Stories, and the Meta AI app. The rollout, announced on Tuesday, follows the release of Muse Spark three months earlier, a large language model that replaced Meta's Llama series. Unlike competitors that release image models to developers first, Meta made Muse Image available to nearly 3 billion users on day one through its existing apps. The model supports complex prompt understanding, real-time web context, and visual reasoning, enabling users to generate images from scratch, edit photos, remove unwanted objects, or create infographics with embedded text.
Muse Image allows conversational, iterative editing. Users can annotate an image with a markup tool, refine outputs step by step, and switch styles or add elements without restarting. A new presets panel offers one-tap options such as restoring old photos, trying trending hairstyles, or transforming selfies into claymation or 16-bit video game visuals. A distinctive feature lets users @-mention Instagram accounts in the Meta AI app to pull in public profile photos for image generation, a function that introduces potential consent concerns. Meta did not release a standalone product or public API alongside this launch, choosing instead to integrate the model natively into its ecosystem.
The move cuts Meta's reliance on third-party image models like those from Midjourney and Black Forest Labs, which previously powered its AI features. Internally, Meta benchmarks show Muse Image trailing OpenAI's GPT Image 2 but outperforming Google's Nano Banana 2 in single and multiple image editing tasks. The model was developed under the codename Mango within MSL, led by Alexandr Wang. A video-generation model, Muse Video, is expected soon. Meta says all images created with Muse Image carry embedded metadata to indicate AI origin, though details on enforcement were not provided.
Meta is prioritizing mass reach over technical dominance by launching Muse Image directly into apps used by billions, bypassing developer-first norms. The integration into WhatsApp and Instagram gives it immediate scale, but the model's performance still lags behind OpenAI's latest. Embedding AI deeply into social interactions raises new questions about consent, especially with features that pull in others' public photos. The lack of a public API at launch suggests Meta is keeping control tight as it tests user response.
💡 NaijaBuzz Take is AI-assisted editorial opinion, not established fact. Full disclaimer →
