Live, steerable video models
These models keep generating while you steer them: a new prompt, a different source video or a changed scene takes effect without starting a new clip.
They are not navigable worlds. You direct what happens on screen rather than walking through a place.
7 models · Alphabetical, never ranked · Updated 2026-11-03
Live video
EdenLive
VAST
Source video (game footage or a live camera), a scene prompt and a reference image
Custom termsOpen weightsReal-time claimTry on Reactor, Windflow
Live video
Helios
PKU-YuanGroup
Text prompts that can change while the video streams
Permissive licenseOpen weightsReal-time claimTry on Hugging Face Space, Reactor, Windflow
Live video
Krea Realtime 14B
Krea
Text prompts and live video input
Permissive licenseOpen weightsReal-time claimLive video
LongLive 2.0
NVIDIA (NVlabs)
Text prompts switched during generation, across multiple shots
Custom termsOpen weightsReal-time claimTry on Reactor, Windflow
Live video
PixVerse R1
PixVerse
Live prompts and chat in persistent, shared worlds
Service termsNo public weightsReal-time claimTry on the official site
Live video
Visko Orbis
Visko
Text prompts and an optional start image, steered between generated chunks
Service termsNo public weightsReal-time claimTry on Reactor
Live video
Vivix W1
Vivix
Text, voice and motion
Service termsNo public weightsReal-time claimTry on the official API and playground
Every model listed here has its own page, a resolved license reading and at least one way to run or try it. Models that do not clear that bar stay in the full directory until their sources do.
The data is open and accepts corrections. View the dataset on GitHub