Luma AI Launches Text-to-Video Model Ray3 with Built-in Reasoning

PALO ALTO, Calif., September 18, 2025 –Artificial intelligence startup Luma AI Inc. has unveiled Ray3, its most advanced text-to-video model to date, designed for high-quality cinematic production. The model introduces chain-of-thought reasoning, enabling it to plan scenes, refine outputs, and collaborate with users more intelligently than existing generative video systems.

“Ray3 is our first step toward building intelligence for creative work,” said Amit Jain, co-founder and CEO of Luma AI. “Creative work is one of the most intellectually challenging things humans do, yet until now, much of the AI available to creatives has lagged far behind what’s possible in coding and analysis with language models.”

Ray3’s reasoning ability allows it to “think” through instructions, sketch storyboards, and respond to user annotations, giving filmmakers and advertisers unprecedented control.

" More than twice the size of Ray2, Ray3 delivers new levels of fidelity, instruction following, and temporal coherence — and it’s the first model capable of generating High Dynamic Range video in ACES EXR, making it purpose-fit for high-end production pipelines. This is the leap forward that filmmakers, advertisers, and game developers have been waiting for," Jain added.

Luma announced a partnership with Adobe to integrate Ray3 into its Firefly creative suite. Launch partners include Dentsu Digital, Monks, StrawberryFrog, and Saudi-based Humain, which praised Ray3 as a “quantum leap” in AI-driven creative tools.

"As Luma AI’s first launch partner for Ray3, Firefly customers get early access to the model ahead of its broader public release. For the next two weeks, Ray3 will be available only in our Firefly app and on Luma AI’s Dream Machine platform," Adobe said.