Video Runtime
Developer-onlyClip-AheadOfficial source

CausalWM

A world model that reasons through motion and geometry with an explicit chain of thought before predicting future video.

Access and release status are source-checked descriptions, not a live uptime check. A reachable website does not establish that video playback or interaction currently works.

Open Documentation ↗View GitHub ↗

No rights-cleared product preview yet. See the official page for its latest interface.

Published by Video Runtime · Source check 2026-09-21 · Evidence method · Suggest a correction

EDITORIALLY VERIFIED PRODUCT INFORMATION

About CausalWM

A world model that reasons through motion and geometry with an explicit chain of thought before predicting future video. Given a first-frame image and an instruction, CausalWM generates an explicit causal chain — optical flow for motion, XYZ pointmaps for geometry — and a shared diffusion transformer turns that reasoning into future RGB video with stage-causal attention.

Key features

Releases TI2V chain-of-thought inference code and model weights on Hugging Face. The intermediate flow and pointmap predictions make the reasoning path inspectable rather than implicit.

Getting started

Download the CausalWMv1 weights from Hugging Face, install the documented environment, then run the TI2V chain-of-thought inference example from the repository quickstart.

Real-time interaction

Given a first-frame image and an instruction, CausalWM generates an explicit causal chain — optical flow for motion, XYZ pointmaps for geometry — and a shared diffusion transformer turns that reasoning into future RGB video with stage-causal attention.

Pricing

LTX-2 Community License code; self-hosted compute costs extra Repository code is published under LTX-2 Community License. Confirm separate model, dataset and dependency terms before reuse. No hosted price is inferred. Budget for the documented GPU, storage and setup requirements.

Limitations

The repository uses the LTX-2 Community License rather than a standard permissive open-source licence. The release targets text-and-image-conditioned future-video prediction; no real-time streaming profile is documented.

Read the documentation ↗

EDITORIAL VIEW

Editor's Verdict

CausalWM is included as a source-available implementation relevant to real-time generative video or interactive world systems.

Run the supplied example on representative hardware and review every linked licence before production use.

How It Works

  • Given a first-frame image and an instruction, CausalWM generates an explicit causal chain — optical flow for motion, XYZ pointmaps for geometry — and a shared diffusion transformer turns that reasoning into future RGB video with stage-causal attention.

What We Like

  • Releases TI2V chain-of-thought inference code and model weights on Hugging Face.
  • The intermediate flow and pointmap predictions make the reasoning path inspectable rather than implicit.

Current Limitations

  • The repository uses the LTX-2 Community License rather than a standard permissive open-source licence.
  • The release targets text-and-image-conditioned future-video prediction; no real-time streaming profile is documented.

Pricing & Access

  • Repository code is published under LTX-2 Community License. Confirm separate model, dataset and dependency terms before reuse.
  • No hosted price is inferred. Budget for the documented GPU, storage and setup requirements.

Verification Summary

  • Official repository and licence reviewed on 2026-09-21. Video Runtime has not installed or benchmarked this project.
  • Performance and compatibility statements below are attributed to the maintainers and are not site measurements.
Technical notes & official performance

Official Performance Data

  • The README describes the causal chain-of-thought design but publishes no FPS, latency or benchmark tables in the inspected sections. Video Runtime has not run it.

Technical Notes

  • The September 2026 release is CausalWMv1 with inference code and weights; the technical report is on OpenReview.

Sources

  1. Official GitHub repository ↗github · accessed 2026-09-21
  2. Code licence ↗github · accessed 2026-09-21
  3. Official project page ↗official-docs · accessed 2026-09-21

Access, stage & mechanism

These are separate source-checked facts, not a live uptime monitor. Unverified fields are left open; transport and persistent sessions do not establish how frames are generated.

Release stage
Not verified
Access mode
Not verified
Source availability
Not verified
Transport
Not verified
Interaction
Not verified
Session mode
Not verified
Generation mechanism
Not verified
Continuity method
Not verified
Licence status
Not verified
Tasks
Not classified

Read how official-source and hands-on records stay separate in our Methodology.