# Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform

arXiv: https://arxiv.org/abs/2605.23972 · HTML: https://mlsanity.com/paper/2605.23972/

## 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 17 of its 36 references that are in the map.
- First posted on arXiv 13 May 2026.

## Placed in

- World models / [World models for language agents](https://mlsanity.com/wm/world-models-for-language-agents/)
- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 17 of 17)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [World Models](https://mlsanity.com/paper/1803.10122/) (2018)
- [Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models](https://mlsanity.com/paper/2310.04406/) (2023)
- [Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task](https://mlsanity.com/paper/2210.13382/) (2022)
- [The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity](https://mlsanity.com/paper/2506.06941/) (2025)
- [Large Language Model Guided Tree-of-Thought](https://mlsanity.com/paper/2305.08291/) (2023)
- [The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs](https://mlsanity.com/paper/2509.09677/) (2025)
- [Emergent World Models and Latent Variable Estimation in Chess-Playing Language Models](https://mlsanity.com/paper/2403.15498/) (2024)
- [Large Language Models Are Not Strong Abstract Reasoners](https://mlsanity.com/paper/2305.19555/) (2023)
- [On the Out-Of-Distribution Generalization of Multimodal Large Language Models](https://mlsanity.com/paper/2402.06599/) (2024)
- [CounterBench: Evaluating and Improving Counterfactual Reasoning in Large Language Models](https://mlsanity.com/paper/2502.11008/) (2025)
- [Failure Modes in LLM Systems: A System-Level Taxonomy for Reliable AI Applications](https://mlsanity.com/paper/2511.19933/) (2025)
- [Evaluating Interventional Reasoning Capabilities of Large Language Models](https://mlsanity.com/paper/2404.05545/) (2024)
- [Dream and Search to Control: Latent Space Planning for Continuous Control](https://mlsanity.com/paper/2010.09832/) (2020)
- [Reasoning Capabilities of Large Language Models on Dynamic Tasks](https://mlsanity.com/paper/2505.10543/) (2025)
- [Learning Local Causal World Models with State Space Models and Attention](https://mlsanity.com/paper/2505.02074/) (2025)
- [Feedback-Induced Performance Decline in LLM-Based Decision-Making](https://mlsanity.com/paper/2507.14906/) (2025)

## Abstract (arXiv)

Large language models achieve strong performance in language generation and knowledge-intensive tasks, yet remain limited in settings requiring causal reasoning, persistent state tracking, and long-horizon planning. We argue that these limitations may arise from an objective-level mismatch between sequence prediction and reasoning over latent environment dynamics. To formalize this distinction, we introduce Latent Dynamics Inference (LDI), a conceptual perspective that interprets language and multimodal observations as partial evidence of underlying transition dynamics. To empirically investigate this perspective, we introduce Flux, a sequential reasoning environment specified entirely through natural-language rules. As a proof-of-concept case study, the rules are first compiled into an explicit state-transition simulator, illustrating that structured latent transition dynamics can, in some cases, be operationally extracted from textual rule descriptions. This enables a controlled comparison between the LLMs operating purely over textual observations and reinforcement-learning agents trained directly within the extracted latent state space. Within this case study, agents operating with explicit access to the latent state space exhibit substantially more stable behavior in long-horizon gameplay, achieving an aggregate win rate of approximately 79% versus 11% for LLMs. Qualitative analysis further reveals failure modes consistent with unstable persistent state tracking, including invalid actions, state-tracking errors, and short-horizon reasoning failures. The complete implementation of the Flux environment available at https://github.com/FeisalAlaswad/FLUX-RL-Agent Within the evaluated setting, these results suggest that strong sequence prediction alone may struggle to support robust long-horizon dynamic reasoning without mechanisms for persistent state tracking and transition modeling