TheFluxTrain is building a world model for film production
The model takes care of continuity and consistency — who is in the frame, what just happened, what happens next — so filmmakers can move from a script to production. Arrival-01 is the first model we are working on. It generates the next frame, then the next, at sub-second steps.
Why we exist
A script is already a plan. Production should follow it.
We spent a year stacking the production tools: LoRA training, storybooks, then video, voice, Flow Studio, and an agentic film pipeline. Once those were running, the gap was obvious. You can generate a shot. You still cannot hold a scene.
Video models invent isolated clips. A film needs continuity and consistency — the same people, the same place, the next frame following the last. That is what a world model is for. Filmmakers should be able to take a script and go to production, not gamble on twenty-second takes.
Two layers, one company
Studio, shipping
Write a story, run an agentic workflow, generate scenes, and finish on a timeline. That platform is live.
Arrival-01, in progress
Our first world model. It generates the next frame, then the next, at sub-second steps, and holds continuity and consistency so a script can go straight into production.
Timeline
How we got here
From image LoRA training in 2024 to an agentic film platform in 2026. Arrival-01 starts from what that platform made obvious.
August 2026
NowArrival-01
Our first world model for film production. It is being built to hold continuity and consistency, generate the next frame then the next at sub-second steps, and let filmmakers move from a script to production.
June–July 2026
Agentic film platform
We completed the agentic platform for film production: story to scenes, generation, and a timeline you can actually edit.
February 2026
Agentic flows
We started building agentic flows and Flow Studio workflows so a film could be planned and run as a graph, not a pile of one-off generations.
November 2025
Filmmaking stack
We moved into filmmaking and added the video, image, and voice models needed to produce shots, not only stills.
August 2025
Omni Context model
We released our internal image Omni Context model so generations could follow a subject across poses, products, and layouts.
Omini Kontext researchJanuary 2025
Storytelling and storybooks
We shifted into storybook creation: characters, scenes, and illustrated stories, not only single-image generation.
September 2024
Image LoRA training
We started as a platform for training image LoRA models so people could keep a face, product, or character consistent without a research setup.
Use the studio today
Arrival-01 is underway. The film platform is live.