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

InSpatio-WorldFM: An Open-Source Real-Time Generative Frame Model

arXiv 26.3 2026 48.9 method

TLDR

Open-source real-time generative frame model for spatial intelligence, achieving low-latency multi-view consistent world simulation on consumer GPUs.

Reasoning

Strengths include a novel frame-based paradigm that reduces latency compared to video-based models, multi-view consistency via 3D anchors and spatial memory, and a progressive training pipeline. Weaknesses are the lack of detailed experimental comparisons and limited information on evaluation metrics in the abstract.

Read-first score

Read-first score 48.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 47.

Recency 6%
100

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

Topical relevance 29%
67.1

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%
60

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=experiment,result

Reproducibility 18%
38

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

Citation impact 18%
33.8

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.33777937

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: 302.

Keyword Scores

world model
9
generative world model
9
interactive world model
9
world simulator
8
video world model
7
world dynamics prediction
3
model-based reinforcement learning world model
2

Deep Analysis

Innovations

  • Frame-based paradigm for real-time spatial inference, generating each frame independently to reduce latency compared to video-based world models
  • Enforcing multi-view spatial consistency through explicit 3D anchors and implicit spatial memory
  • Progressive three-stage training pipeline: pretrained image diffusion model -> controllable frame model -> real-time generator via few-step distillation

Methodology

InSpatio-WorldFM adopts a frame-based paradigm that generates each frame independently, using explicit 3D anchors and implicit spatial memory to enforce multi-view consistency. The model is trained via a progressive three-stage pipeline that transforms a pretrained image diffusion model into a controllable frame model and then into a real-time generator through few-step distillation.

Key Results

The model achieves strong multi-view consistency and supports interactive exploration on consumer-grade GPUs, providing an efficient alternative to traditional video-based world models for real-time world simulation.

Tags