Luma AI

Luma AI (Luma Labs) is a creative platform where agents research, generate, and refine work across video, image, audio, and text. It is built on models the company develops itself, including Ray 3.2 for video and Uni-1 for image generation with brand grounding. Luma also offers an API and an enterprise offering.

What It Is

Luma is a multimodal creative workspace organized around boards, where an agent helps users brainstorm, research, and create. According to the site, users can generate many variations at once, edit existing images and videos, animate images, add voiceovers, lip sync and captions, and collaborate on shared boards. Learning Center guides cover layers editing, Luma Scenes (turning a brief into an editable storyboard and finished video), character consistency, upscaling, HDR/EXR color spaces, Luma Connectors, and Luma Skills for repeatable workflows.

Models and Research

Luma states that it builds the models underneath its product. Ray 3.2 is described as giving control over every frame, continuity across cuts, and cinematic-quality finishing. Uni-1 is described as carrying style references, character references, and visual grounding into assets. The site also lists third-party models in its guides, such as Seedance, FLUX, and GPT Image. Luma also announces an Open Physical AI Lab and publishes engineering research posts.

Enterprise Use

The enterprise page describes image, video, and sound production in one place, SSO and access controls, and dedicated Forward Deployed Creatives for onboarding. Luma states that content created is not used to train models. Listed workflows include campaign variations, product visuals, social video ads, storyboards, packaging mockups, and video localization.