Published by Video Runtime · Source check 2026-09-14 · Evidence method · Suggest a correction
EDITORIALLY VERIFIED PRODUCT INFORMATION
About ForgeWM
Train playable action-conditioned video world models with an open progressive causal recipe. Applies four progressive stages from bidirectional fine-tuning through on-policy distillation to create one-, two- and four-step students.
Key features
Releases training and inference code, checkpoints and pre-encoded data with keyboard, mouse and gamepad control examples.
Getting started
Open the official repository, read its current README and licence, then follow the documented environment, checkpoint and inference steps. Start with the supplied example before using your own media or data.
Real-time interaction
Applies four progressive stages from bidirectional fine-tuning through on-policy distillation to create one-, two- and four-step students.
Pricing
Apache-2.0; compute costs extra Repository code: Apache-2.0. Model weights and dependencies may have separate terms. No hosted price is estimated here. Budget for your own GPU or cloud usage after testing a representative workload.
Limitations
The full training recipe is documented for eight GPUs, and the released data and model dependencies require significant storage and compute.
EDITORIAL VIEW
Editor's Verdict
Consider ForgeWM if its documented workflow matches your experiment; start from the linked repository rather than an unofficial mirror.
Check the licence and run a small local evaluation before choosing it for production. No ratings or measured results are inferred from its publication.
How It Works
- Applies four progressive stages from bidirectional fine-tuning through on-policy distillation to create one-, two- and four-step students.
What We Like
- Releases training and inference code, checkpoints and pre-encoded data with keyboard, mouse and gamepad control examples.
Current Limitations
- The full training recipe is documented for eight GPUs, and the released data and model dependencies require significant storage and compute.
Pricing & Access
- Repository code: Apache-2.0. Model weights and dependencies may have separate terms.
- No hosted price is estimated here. Budget for your own GPU or cloud usage after testing a representative workload.
Verification Summary
- Official repository and licence reviewed on 2026-09-14. Video Runtime has not installed or benchmarked this project.
- This is a source review, not verification of demo uptime, output quality or performance on your hardware.
Technical notes & official performance
Official Performance Data
- The authors report up to 72 FPS for the one-step model at 352×640 on one H20. This is not an independent site benchmark.
Technical Notes
- The primary model uses a game-native image-to-video backbone, with a separate CrossFPS lineage for gamepad-driven experiments.
Sources
- Official GitHub repository ↗github · accessed 2026-09-14
- Code licence ↗github · accessed 2026-09-14
- Official project documentation ↗official-docs · accessed 2026-09-14
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
CONTINUE EXPLORING
Related projects & models
Selected by research links and category, not by popularity.
Generate continuing interactive worlds with bounded external state, mid-rollout prompts and camera control.
Another frame streaming option in this section.Public repository; self-hosted setup and compatible GPU required
Train few-step autoregressive video generators with frame-wise and chunk-wise Causal Forcing variants.
Another frame streaming option in this section.Public repository; self-hosted setup and compatible GPU required
Learn and reproduce the full pipeline for converting video diffusion backbones into interactive world models.
Another frame streaming option in this section.Public repository; self-hosted setup and compatible GPU required
Run an action-conditioned interactive world model with open-ended rollout on a single desktop GPU.
Another frame streaming option in this section.Public repository; self-hosted installation and compatible GPU requiredRead how official-source and hands-on records stay separate in our Methodology.