Mareel

Mareel is an all-in-one AI video and image generation platform that brings 40+ leading AI models into one workspace. Instead of opening a new account every time a new model launches, you write one brief, pick the model that fits the shot, and get a finished video or image back.

New video and image models ship every few weeks, and each one is strong at something different. One handles realistic human motion, another renders accurate text inside images, another generates native dialogue and sound. Keeping up usually means juggling subscriptions, credit systems, and file folders across a dozen sites. Mareel exists so that keeping up with them is not a project of its own: one account, one credit balance, and one place where everything you make ends up.

What you can do with Mareel:

Generate AI videos from text, from a single image, from first and last frames, or from multiple reference images, depending on the model Create and edit images, with style templates and support for multiple reference images on supported models Produce ad creatives in Ad Studio: UGC-style ads, remakes of proven ad structures around your own product, product-link-to-video ads, product presenter videos, and complete product photo sets Make talking-photo videos, swap the lead character in a video, transfer motion from one clip to a character, and generate voiceovers Plan a short film on Canvas: break a story into shots, generate every frame and clip, and assemble the cut Start from Marketing Lab templates: pick a proven recipe, drop in your product image, and generate Keep reusable product and character libraries so your product and cast look the same in every output

Every submission is checked against the selected model before it is accepted. Reference-image counts, generation mode, resolution, duration, and prompt length all have to be supported, so a mismatch is refused instead of charged. Credits are reserved when a job starts and returned in full if it fails, and the credit cost is shown before you generate. Video jobs run on a queue, so closing the tab does not stop them, and every output lands in your Library with its status, settings, and the model that produced it.