⌁ Awesome World Model Hub Papers · Datasets · Projects
中文 GitHub
← Back to papers

Infinite Worlds with Versatile Interactions

arXiv 2026 43.9 method, system, application

TLDR

LingBot-World 2.0 achieves unbounded interaction horizon, real-time 720p 60fps video, diverse actions, and agentic harness for world simulation.

Reasoning

The paper presents impressive technical advances in world simulation, including unbounded interaction, real-time video, diverse actions, and an agentic harness. However, it lacks empirical evaluation or comparison to baselines, and no real-world benchmarks are mentioned.

Read-first score

Read-first score 43.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 57.

Recency 6%
100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 29%
81.4

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 18%
50

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=none

Reproducibility 18%
30

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation impact 18%
0

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 12%
0

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 473.

Keyword Scores

world model
10
world simulator
10
interactive world model
10
generative world model
9
video world model
9
world dynamics prediction
7
model-based reinforcement learning world model
2

Deep Analysis

Innovations

  • Unbounded interaction horizon with consistent output quality via causal pretraining paradigm
  • Real-time distilled variant capable of 720p video at 60 fps
  • Highly diverse interactive elements including broader actions (attacking, archery, spell-casting, shooting) and text-driven events
  • Integration of agentic harness with pilot agent for character behavior planning and execution, and director agent for environmental synthesis
  • Multi-player shared experience interface
  • Lightweight 1.3B counterpart for single-GPU deployment

Methodology

The system builds on a causal pretraining paradigm to achieve an unbounded interaction horizon, distills a real-time variant for fast inference, and incorporates an agentic harness with a pilot agent (plans character behaviors) and a director agent (synthesizes novel environmental elements). A multi-player interface is provided, and the primary 14B model is paired with a 1.3B lightweight model for efficient deployment.

Key Results

The model achieves an unbounded interaction horizon without quality degradation, sustains real-time 720p video generation at 60 fps via the distilled variant, and supports diverse action types and agentic world modeling.

Limitations

  • No explicit limitations are discussed in the abstract

Resources

Paper

Tags