Luma AI

Ray3 video models

The latest Ray3.14 generates native 1080p video with accurate motion and physics, around 4x faster and at lower cost than earlier versions.

Text-to-video and image-to-video

Create clips from a written prompt or animate a still image, with keyframe control over start and end frames.

Modify Video (Ray3 Modify)

Restyle and edit existing footage, swap scenes and guide AI performances while keeping continuity, aimed at film and post-production.

Luma Photon image generation

Photon and Photon Flash produce photorealistic stills, with Flash tuned for fast, cheaper iteration.

Luma Agents

Agents plan, generate and refine across a project, turning one brief into multiple shots and edits with shared context.

Multi-model workspace

One app routes to Luma's models plus third-party options such as Veo, Kling and GPT Image, and ElevenLabs for audio.

UNI-1 multimodal model

Luma's unified understanding-and-generation model underpins prompting and is available through the API.

Character and consistency controls

Character Reference and keyframes keep the same subject and look across separate shots.

Reframe and aspect-ratio tools

Convert footage between horizontal and vertical and adapt clips for different platforms.

Voice input and project boards

Turn spoken ideas into prompts, and assemble generations into boards with captions and EXR export for professional pipelines.

Developer API

The Luma API, including UNI-1.1, lets teams build video and image generation into their own apps and workflows.