# OpenWAM: An Open, Modular Exploration Towards Systematic World-Action Model Pretraining

arXiv: https://arxiv.org/abs/2609.07398 · HTML: https://mlsanity.com/paper/2609.07398/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Placed in Embodied and robot world models, one of 11 directions in World models.
- First posted on arXiv 7 Sep 2026.

## Placed in

- World models / [Embodied and robot world models](https://mlsanity.com/wm/embodied-and-robot-world-models/)

## Built on by (most cited first, 22 of 22)

- [MachEmbodied-U0: Unified Understanding and Generation Model for Embodied Intelligence](https://mlsanity.com/paper/2609.25627/) (2026)
- [InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data](https://mlsanity.com/paper/2609.31394/) (2026)
- [UniWAM: Unified World-Action Model](https://mlsanity.com/paper/2610.02054/) (2026)
- [RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments](https://mlsanity.com/paper/2610.10409/) (2026)
- [Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement](https://mlsanity.com/paper/2610.12369/) (2026)
- [RoboHarness: A Simple Harness Outperforms VLA and World Actions Models](https://mlsanity.com/paper/2610.14018/) (2026)
- [CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models](https://mlsanity.com/paper/2609.18462/) (2026)
- [SCULPT-VLA: Learning Structured Control through Staged Action Grounding](https://mlsanity.com/paper/2609.23275/) (2026)
- [An Action Is Worth One Patch: Unified World-Action Modeling with PatchWAM](https://mlsanity.com/paper/2609.25961/) (2026)
- [InternW0: A Foundational Physical World Model for Efficient Real-World Interactions](https://mlsanity.com/paper/2609.27656/) (2026)
- [Demonstration-Free Success-Probability Reward Learning for Generalist Robot Policies](https://mlsanity.com/paper/2609.33653/) (2026)
- [PreferenceFlow: Test-Time Guidance of Flow-Matching Robot Policies from Human Interventions](https://mlsanity.com/paper/2609.36872/) (2026)
- [Direct Experience World-Model Optimization: Learning the World Beyond Action Imitation](https://mlsanity.com/paper/2609.37398/) (2026)
- [GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives](https://mlsanity.com/paper/2609.39601/) (2026)
- [Magic-W0: A Structured World-Action Foundation Model for Physical Intelligence](https://mlsanity.com/paper/2609.39870/) (2026)
- [Ego4WAM: What Matters When Scaling Egocentric Human Data for Robot Learning?](https://mlsanity.com/paper/2609.40341/) (2026)
- [WAMJET: A Harness for World Action Model Acceleration](https://mlsanity.com/paper/2610.03797/) (2026)
- [Vela: Scaling Vision-Language-Action Models with Adaptive Action Curve Parametrization](https://mlsanity.com/paper/2610.05230/) (2026)
- [OpenWAM: An Open Framework for Composable World-Action Models](https://mlsanity.com/paper/2610.07922/) (2026)
- [Video Prediction Policy 2: Predict Better, Act Better](https://mlsanity.com/paper/2610.10270/) (2026)
- [WAM-Cache: Staleness-Bounded KV Reuse for Efficient World Action Models](https://mlsanity.com/paper/2610.11401/) (2026)
- [BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries](https://mlsanity.com/paper/2610.14746/) (2026)

## Abstract (arXiv)

World-Action Models inherit world knowledge from video-generative priors, and channel it into executable control signals through embodied experience. Existing systems, however, are monolithic: the generative backbone, visual representation, architecture, information flow, inference procedure, and training data are tightly coupled, obscuring which design choices matter and why. We introduce OpenWAM, an open research stack that turns world-action pretraining into a controlled experimental program. OpenWAM-Infra factorizes the WAM design space into composable modules with unified training, inference, deployment, and evaluation. On this substrate, OpenWAM-Study examines three questions through controlled experiments: what to inherit, how world and action learning interact, and how their synergy scales; and distills three principles: upstream knowledge transfers through a sufficiently capable generative backbone and a compact, information-rich latent space; world-action synergy requires dedicated action capacity, explicit world-to-action information flow, and synchronized joint denoising; and embodied pretraining principally improves out-of-domain generalization, with one-stage co-training over egocentric and robot data integrating world coverage and action grounding. Composing these principles, we build OpenWAM-α, an open WAM pretrained on roughly 6,400 hours of egocentric human and robot data and evaluated across simulation and real-world benchmarks. Across the eight simulation benchmarks and the real-robot experiments, which together span embodiments from single-arm and bimanual manipulation to dexterous hands, OpenWAM-α delivers consistently excellent performance, sustaining its top-tier standing from simulation to the physical world. We release the full stack, including infrastructure, evaluation protocols, pretrained models, and data recipes, to facilitate future research.