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

ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPU

arXiv 2026 47 method, system

TLDR

ABot-World-0 is a real-time action-conditioned video world model for long-horizon interaction, running on a single desktop GPU.

Reasoning

The paper presents a novel approach for real-time interactive video generation with long-horizon consistency using distillation and LongForcing. However, its reliance on specific data sources and infrastructure may limit generalizability, and the evaluation is limited to a few benchmarks.

Read-first score

Read-first score 47, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 61.

Recency 6%
100

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

Topical relevance 29%
87.1

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=experiment

Reproducibility 18%
38

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code

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: 579.

Keyword Scores

world model
10
interactive world model
10
video world model
10
world dynamics prediction
10
generative world model
9
world simulator
8
model-based reinforcement learning world model
4

Deep Analysis

Innovations

  • Action-conditioned video world model for real-time, long-horizon closed-loop interaction on a single desktop GPU.
  • WorldExplorer agent-driven data collection guided by training feedback.
  • Unified annotation pipeline with 14 deterministic quality checks and VLM-based assessment for synchronized action and text labels.
  • Progressive distillation from bidirectional teacher to causal student using teacher forcing, ODE distillation, and LongForcing to mitigate autoregressive drift.
  • Unified keyboard action interface for scene roaming and third-person character interaction.
  • Reference-character memory for persistent identity consistency during long rollouts.
  • Co-designed streaming inference stack with lightweight VAE, efficient attention, memory-aware scheduling, and low-bit DiT inference.

Methodology

The model is trained on multi-source data from AAA games, simulation engines, and internet videos. WorldExplorer performs agent-driven data collection informed by training feedback, and a unified pipeline applies deterministic quality checks, VLM-based assessment, and synchronized action-text annotation. A bidirectional action-conditioned teacher is progressively distilled into a causal student via teacher forcing, ODE distillation, and the novel LongForcing technique to align long self-rollouts, while raw keyboard actions and reference-character memory ensure controllable interaction and identity consistency. Deployment uses a streaming inference stack co-designed with a lightweight VAE decoder, efficient attention, memory-aware scheduling, and low-bit DiT.

Key Results

ABot-World-0 streams 720P video at up to 16 FPS on a single NVIDIA RTX 5090 with 1.2s action-to-first-frame latency and ~19GiB peak VRAM, demonstrating competitive controllability and coherent long-horizon evolution on WorldRoamBench and extended interactive rollouts.

Resources

Paper

Tags