Long-WAM: Scaling the Context of World-Action Models

Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action.

By Huang, Zhang, Liu +13arXivOct 7, 2026

Score█████░░░░░5.4

Key numbers

63.3% to 78.7%
95% success on dynamic cup

Caveats

Preprint; not yet peer reviewed.

VerdictCompetent work. Briefs at most.

Read the original PDF Code

Abstract

Real-time robot control demands enough visual history to infer motion and task progress, but processing that history can delay action. We present Long-WAM, a model-system framework for scaling the context of causal world-action models under real-time control constraints. Our central finding is that access to history is not the same as using it: longer histories pay off far more when the video foundation is pretrained autoregressively (AR). We first learn causal prediction from robot and egocentric videos without action labels, then preserve this history-to-future structure during world-action adaptation. On RoboCasa GR-1, increasing context from 0.0 to 19.2 seconds raises success from 63.3% to 78.7%, whereas a bidirectionally pretrained initialization shows no net gain; robot-domain AR pretraining further raises peak success on GR-1 and LIBERO-Long. Long-WAM also achieves the best results among compared methods on LIBERO-Long, RoboTwin 2.0, and DOMINO. Streaming observation encoding, asynchronous execution, and hardware-specific acceleration enable deployment on RTX 5090, DGX Spark, and Jetson AGX Thor without dropping future prediction; on RTX 5090, each action chunk, including future-video latent prediction, takes 107.4 ms. Real-time deployment on Unitree G1 and YAM supports dynamic and long-horizon manipulation, including 95% success on dynamic cup stacking, where Pi0.5 and Fast-WAM succeed in none of 20 trials. As a memory-informed executor, Long-WAM also complements higher-level planning in composite tasks.

Wei Huang, Bohan Zhang, Chenzhi Liu, Isabella Liu, Shuai Yang, Weian Mao, Luozhou Wang, Yicheng Xiao, Weifeng Lin, Qixin Hu, Bryan Chu, Sifei Liu, Linxi Fan, Xiaojuan Qi, Song Han, Yukang Chen

The editor's rubric

Heuristic review

Dimension	Level	Weight	What that level means
Leverage	███░░ 3	24%	A method or resource many groups across the field will adopt within a year.
Magnitude	███░░ 3	18%	Large gain: roughly 2x, or a clear new state of the art on a hard, unsaturated problem.
Evidence	██░░░ 2	14%	Limited: single setting, weak baselines, or an observational association presented as causal.
Novelty	██░░░ 2	16%	A new combination of known ideas.
Trajectory	██░░░ 2	18%	Some room to improve with obvious engineering.
Stakes	██░░░ 2	10%	Benefits a professional community (practitioners, clinicians, engineers).

Editor’s rationale

Heuristic triage from title and abstract text only, not a reading of the paper. Cues found: method (we propose); gains (relative gain); scale (scalable).

How the score was computed

rank-2026-10-07

Score█████░░░░░5.4

Score = 10 × (65% × adjusted merit / 10 + 25% × attention + 10% × freshness)

Merit: 4.0 / 10; Weighted rubric, evidence-gated.
Adjusted merit: 4.0 / 10; Shrunk toward the desk prior by editor confidence (36%).
Attention: 84%; Citations, upvotes, points, mentions.
Freshness: 73%; Half-life decay since publication.

Hugging Face upvotes27 (reference 25, via hf-daily, Oct 8, 2026, 03:47 UTC)
GitHub stars2.7k (reference 250, via hf-daily, Oct 8, 2026, 03:47 UTC)

The record

Reviewed by heuristic-v5 on Oct 8, 2026, 03:46 UTC. Paper type: method.
TOP, No.3 in the Artificial Intelligence edition of October 8, 2026.