# Beyond Visual Quality: A Study of Test-Time Planning with World Action Models

arXiv: https://arxiv.org/abs/2609.24745 · HTML: https://mlsanity.com/paper/2609.24745/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in World models for language agents, one of 11 directions in World models.
- Builds on 25 of its 47 references that are in the map.
- First posted on arXiv 21 Sep 2026.

## Placed in

- World models / [World models for language agents](https://mlsanity.com/wm/world-models-for-language-agents/)

## Builds on (most cited first, 25 of 25)

- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [World Action Models are Zero-shot Policies](https://mlsanity.com/paper/2602.15922/) (2026)
- [WorldVLA: Towards Autoregressive Action World Model](https://mlsanity.com/paper/2506.21539/) (2025)
- [Fast-WAM: Do World Action Models Need Test-time Future Imagination?](https://mlsanity.com/paper/2603.16666/) (2026)
- [Causal World Modeling for Robot Control](https://mlsanity.com/paper/2601.21998/) (2026)
- [Visual Foresight: Model-Based Deep Reinforcement Learning for Vision-Based Robotic Control](https://mlsanity.com/paper/1812.00568/) (2018)
- [WorldArena: A Unified Benchmark for Evaluating Perception and Functional Utility of Embodied World Models](https://mlsanity.com/paper/2602.08971/) (2026)
- [RynnVLA-002: A Unified Vision-Language-Action and World Model](https://mlsanity.com/paper/2511.17502/) (2025)
- [From Foresight to Forethought: VLM-In-the-Loop Policy Steering via Latent Alignment](https://mlsanity.com/paper/2502.01828/) (2025)
- [Inference-time Physics Alignment of Video Generative Models with Latent World Models](https://mlsanity.com/paper/2601.10553/) (2026)
- [τ0-WM: A Unified Video-Action World Model for Robotic Manipulation](https://mlsanity.com/paper/2606.01027/) (2026)
- [Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment](https://mlsanity.com/paper/2602.12281/) (2026)
- [RoboWM-Bench: A Benchmark for Evaluating World Models in Robotic Manipulation](https://mlsanity.com/paper/2604.19092/) (2026)
- [World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry](https://mlsanity.com/paper/2604.01985/) (2026)
- [Is the Future Compatible? Diagnosing Dynamic Consistency in World Action Models](https://mlsanity.com/paper/2605.07514/) (2026)
- [Making Foresight Actionable: Repurposing Representation Alignment in World Action Models](https://mlsanity.com/paper/2606.12217/) (2026)
- [HarmoWAM: Harmonizing Generalizable and Precise Manipulation via Adaptive World Action Models](https://mlsanity.com/paper/2605.10942/) (2026)
- [MiraBench: Evaluating Action-Conditioned Reliability in Robotic World Models](https://mlsanity.com/paper/2605.29360/) (2026)
- [SelfWAM: A Self-Grounded Unified World Action Model for Fast Robot Control](https://mlsanity.com/paper/2608.00725/) (2026)
- [Test-Time Scaling for World Action Models via Zero-Shot Geometric Evaluation](https://mlsanity.com/paper/2607.17454/) (2026)
- [ACID: Action Consistency via Inverse Dynamics for Planning with World Models](https://mlsanity.com/paper/2607.02403/) (2026)
- [RoboWorld: Fast and Reliable Neural Simulators for Generalist Robot Policy Evaluation](https://mlsanity.com/paper/2607.01060/) (2026)
- [DynamicWAM: Dual-Path Motion Conditioning for World-Action Models in Dynamic Manipulation](https://mlsanity.com/paper/2608.00793/) (2026)
- [Foresight Without Seeing: Latent Futures for World Action Models](https://mlsanity.com/paper/2608.11605/) (2026)
- [Keep the Future, Drop the Rollout: RIFT for World Action Models](https://mlsanity.com/paper/2608.11521/) (2026)

## Abstract (arXiv)

World action models generate actions together with visual predictions of their consequences. These paired outputs create the potential for planning by sampling multiple actions from one state, comparing their imagined outcomes, and choosing the action with the most promising predicted outcome. However, how to use imagined futures to guide action selection remains unclear. We examine this planning potential empirically. First, we estimate an oracle upper bound on selection by choosing the sampled candidate whose realised outcome is best. In a controlled same-state analysis, this choice raises success from 68.9% under uniform random selection to 79.2%. We then test selectors based on visual quality, physical consistency, and task progression as controlled interventions. Some tested selectors yield higher observed success, but the gains are uneven and the matched selectors leave much of the measured opportunity unrecovered. To investigate this gap, we examine whether sampled actions lead to different outcomes, whether these differences are visible in the predictions, and whether a score recognises them. Counterfactual branching from the same states shows that selection opportunity is concentrated in relatively few decisions in the initial candidate sets. Action spread and outcome coverage need not increase together. In a further evaluation across trajectory phases with complete action execution, the tested scores again recover little of the available improvement despite a small gain from learned value. These findings distinguish producing consequential action choices from recognising them in generated futures, motivating the evaluation of WAM predictions through their usefulness for decisions rather than visual quality alone.