Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control

arXiv 2026 40.2 method, application

TLDR

Zing-0.5 is a 5B autoregressive world model enabling playable worlds with real-time joint keyboard and text control, using unified conditioning, event-scale supervision, and low-cost streaming.

Reasoning

The paper introduces a strong interactive world model with novel joint action/text conditioning and efficient real-time streaming, backed by benchmark scores and released artifacts. However, the abstract provides limited detail on baselines, ablations, and generalization beyond navigation cases.

Read-first score

Read-first score 40.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 55.

Recency 5%

100

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

Topical relevance 25%

78.6

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 15%

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

Reproducibility 15%

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

Graph prestige 15%

5.7

PageRank-style prestige over the local citation graph built from available OpenAlex references. local_pagerank=0.0016

Citation impact 15%

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

Citation velocity 10%

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 561.

Keyword Scores

world model

generative world model

interactive world model

world simulator

video world model

world dynamics prediction

model-based reinforcement learning world model

Deep Analysis

Innovations

Unified action and text conditioning combining magnitude-aware keyboard inputs with temporally aligned text instructions and jointly annotated videos to learn navigation and event control within the same sequence.
Event-scale supervision for incremental generation using a segment-level teacher trained on connected multi-prompt videos to supervise a block-level causal student through distribution-matching distillation.
Low-cost real-time interaction combining four-step generation with context-preserving streaming to support 832 x 480 inference at 24 FPS at approximately USD 0.009 per stream-minute.

Methodology

Zing-0.5 is a 5B autoregressive world model that conditions on joint keyboard and text inputs for playable world generation. It uses unified action/text conditioning with jointly annotated videos, event-scale distillation from a segment-level teacher to a block-level causal student, and four-step generation with context-preserving streaming. Evaluation uses 158 WBench Navigation cases with overall and consistency scores.

Key Results

Zing-0.5 achieves an overall score of 81.0 and a consistency score of 88.5 across 158 WBench Navigation cases. It also supports 832 x 480 inference at 24 FPS at an estimated server rental cost of approximately USD 0.009 per stream-minute, with a joint-control demonstration showing text-directed event change during continued navigation without restarting generation.

Resources

Paper