Meta launches Muse Image, Video as AI media race moves deeper into social apps

• Photo from Meta

Meta has launched Muse Image and previewed Muse Video, its latest artificial intelligence models for generating and editing media across its platforms.

The company said the two models are the first media generation systems developed by Meta Superintelligence Labs, marking a deeper push into AI tools for images, videos, creators, and businesses.

Muse Image is now available through the Meta AI app, meta.ai, Instagram Stories in the United States, and WhatsApp in limited countries. It is also coming soon to Facebook. Muse Video, meanwhile, will be rolled out soon to creators and Meta AI.

Meta said Muse Image is its most advanced image generation model so far, designed to follow instructions, edit images with precision, compose visuals from multiple references, and use Instagram for social context.

Unlike traditional text-to-image models that directly convert prompts into visuals, Muse Image works more like an AI agent. Meta said the model can use search and coding tools, refine its own outputs, and improve results by spending more computing power during generation.

This means Muse Image can do more than produce images from simple prompts. It can search the web to ground images in real-time information, write and execute code for more accurate details such as charts or QR codes, and revise its own work when parts of an image need correction.

Meta said the search capability is meant to improve accuracy for prompts that involve current events, real-world facts, or visual references. The coding feature, meanwhile, allows the model to generate more precise elements such as plots, diagrams, QR codes, animated GIFs, websites with embedded images, and interactive visual games.

Muse Image also supports image editing and multi-reference composition. Users can ask the model to change specific parts of an image, restore or improve photos, combine elements from different reference images, or continue refining a visual through multiple editing turns.

The model is also connected with Muse Spark, allowing both systems to share tools and plan together for more complex media generation tasks.

Meta said Muse Image can improve through “self-refinement,” where the model reviews its own output and decides whether to make a small edit, generate a new image, or use another tool to improve accuracy. The company said this behavior emerged during reinforcement learning because it produced better results.

The company also said Muse Image improves when given more time and computing resources during inference. According to Meta, deliberate reasoning and tool use can help the model produce stronger results than simply generating multiple images and selecting the best one.

For video, Meta said Muse Video is built on the same pretraining base as Muse Image and supports native audio. The company said the model is designed for visual fidelity, prompt adherence, and temporal consistency.

However, Meta acknowledged that some areas still need improvement, including audio-video synchronization and physically accurate fast motion.

Meta said Muse Image ranked No. 2 on Arena for text-to-image generation, single-image editing, and multi-image editing based on human preference Elo rankings as of July 5, 2026. Muse Video ranked No. 3 on Arena for text-to-video generation during the same period.

The rollout also comes with a safety and transparency feature called Content Seal, Meta’s invisible watermarking system for AI-generated images.

Meta said images created by Muse Image through the Meta AI app and meta.ai will carry a hidden provenance signal that can remain intact even if the image is cropped, compressed, resized, or screenshotted. The company also plans to bring Content Seal to video.

Meta is also previewing a detection tool that can check whether an image carries a Content Seal watermark, giving users another way to identify whether a visual was made using Meta AI.

The launch strengthens Meta’s position in the fast-growing AI media generation market, where technology companies are racing to build tools that can create, edit, and verify synthetic content across social platforms.

For Meta, Muse Image and Muse Video also deepen the role of AI inside its ecosystem, from Instagram and WhatsApp to Facebook and Meta AI, as the company builds more creative tools for users, creators, and businesses.