What it does

Luma describes its product as Luma Agents, a creative workspace for making and organising images, video, audio and text on an infinite canvas. A multimodal agent works out what you want and chooses a suitable model for each request, or you can pick models and adjust parameters yourself. The page lists Luma's own Ray and Uni-1 models alongside third-party ones such as Veo 3.1, Kling 3.0, Seedance 2.0, GPT Image 2, Nano Banana and ElevenLabs voice, music and sound-effect models. One mode is for exploring ideas and strategy without generating any media, while Create Mode produces the assets. Work sits on a zoomable board where you can place, group and compare items, annotate with notes, arrows and drawings, and share the canvas by link or email with Viewer or Editor permission. Master Reference Assets let you define characters, products, logos and brand elements that the agent keeps consistent across variations, video and localised versions. Editing is covered as well as generation: restyle images, animate stills, extend video, reframe aspect ratios, and add voiceover, music, captions and lip sync. The agent can also search the web, analyse videos and pull information from PDFs, and the board supports semantic search for finding assets by description. Team workspaces share credits, and the page mentions EXR export for professional pipelines. All generation is paid for with monthly credits that reset each billing cycle; unused credits expire, and credits spent on failed generations are returned.

Capabilities

Agent that chooses models for each request, with manual model and parameter control available
Infinite shared canvas with notes, frames, annotations and link or email sharing
Master Reference Assets to keep characters, products and logos consistent
Image, video, audio and text generation plus editing, captions and lip sync
Many variations from a single prompt or asset, generated in parallel
Team workspaces with shared credits

When to use it

You need to produce many variations of a campaign, ad or product visual from one idea
You want one workspace for images, video, voiceover and music instead of several subscriptions
You need characters or brand elements to stay consistent across assets
You want to localise a video into other languages with re-synced speech