Planning in 8 Tokens: A Compact Discrete Tokenizer for Latent World Model

2026-03-05

AI Summary

This work proposes CompACT, a novel action-conditional latent world model that dramatically reduces the computational overhead of planning by compressing each observation frame into only eight discrete tokens—orders of magnitude fewer than the hundreds typically used in conventional approaches. By enabling highly efficient sequence modeling and compact representation, CompACT achieves planning performance on par with existing methods across multiple tasks while accelerating inference by several orders of magnitude. This substantial gain in computational efficiency significantly enhances the feasibility of deploying world models in real-time control scenarios, where low-latency decision-making is critical.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

World models provide a powerful framework for simulating environment dynamics conditioned on actions or instructions, enabling downstream tasks such as action planning or policy learning. Recent approaches leverage world models as learned simulators, but its application to decision-time planning remains computationally prohibitive for real-time control. A key bottleneck lies in latent representations: conventional tokenizers encode each observation into hundreds of tokens, making planning both slow and resource-intensive. To address this, we propose CompACT, a discrete tokenizer that compresses each observation into as few as 8 tokens, drastically reducing computational cost while preserving essential information for planning. An action-conditioned world model that occupies CompACT tokenizer achieves competitive planning performance with orders-of-magnitude faster planning, offering a practical step toward real-world deployment of world models.

Problem

• world models
• decision-time planning
• latent representations
• tokenization
• real-time control

Innovation

• compact tokenizer
• latent world model
• decision-time planning
• discrete representation
• real-time control