Luma has launched Luma Agents, an end-to-end creative AI platform for text, image, video, and audio work, powered by its Unified Intelligence models.

A core differentiator is the ability to generate large sets of variations without constant prompting, letting users steer direction through conversation and enabling end-to-end workflows.

The platform is already being used by major clients such as Publicis Groupe, Serviceplan, Adidas, Mazda, and Humain, signaling strong enterprise adoption.

Luma demonstrated notable efficiency gains, turning a $15 million, year-long ad campaign into multiple localized ads across countries in 40 hours for under $20,000, while maintaining internal quality controls.

Luma plans a gradual access rollout to ensure reliable availability and minimize workflow disruptions now that the API is publicly accessible.

The Uni-1 model, the first in the Unified Intelligence family, is trained across audio, video, image, language, and spatial reasoning, with capabilities to render in pixels and process language, and future support for audio and video outputs.

Luma Agents can plan, generate across media, and coordinate with other AI models, enabling persistent context across assets and iterations and allowing self-critique to refine outputs.