Alaya-EVOKE: From Linear-Scaling Supervision to Endless World
Summary
Evoke is an interactive world model with external persistent memory and a redesigned long-horizon teacher, enabling open-ended video generation with bounded context and low latency. It achieves state-of-the-art performance on WBench while staying competitive on VBench-Long and VBench-2.0.
View Cached Full Text
Cached at: 08/14/26, 03:26 AM
Paper page - Alaya-EVOKE: From Linear-Scaling Supervision to Endless World
Source: https://huggingface.co/papers/2608.13546
Abstract
Evoke is an interactive world model that uses external persistent memory and a redesigned long-horizon teacher to enable responsive, open-ended video generation with bounded context and low latency.
Interactiveworld modelsmust support persistent memory, responsive interaction, and long-horizon generation, yet these requirements place conflicting demands on the model. Maintaining history in thedenoiser contextorkey-value cacheincurs growing cost, forcing a trade-off between session length and retained memory, while low-latency interaction relies on few-step generation whose capabilities are bounded by its teacher. Evoke addresses both limitations by externalizing persistent world state and redesigning the teacher for long-horizon interactive generation. Scene geometry is maintained in an external, camera-indexed world state bank, from which only view-relevant information is retrieved, keeping thedenoiser contextbounded as the session grows. Rather than treating the teacher as a fixed generator, we design it for long-horizon supervision: itssparse attentioncombineschunk-wise grouping, retrieval of selected distant frames, and a linear-attention global state, yielding linear growth in memory and compute while enabling supervision over long horizons. Such supervision exposes content drift that stays locally plausible within short windows, while per-chunk conditioning enables prompt changes and event control throughout the sequence. A 30-seconddistribution-matching objective, applied underself-forced rollouts, transfers both capabilities to a three-step student that uses noclassifier-free guidance, improving resistance to long-term drift while preserving responsive conditioning. With bounded context and recurrent external memory, Evoke supports open-ended, continuously evolving generation; on a single H200 at 384times 640, each 1.5,s chunk is generated in 2.11,s. As a three-step world model, Evoke achieves state-of-the-art performance on WBench while remaining competitive on VBench-Long and VBench-2.0.
View arXiv pageView PDFProject pageAdd to collection
Get this paper in your agent:
hf papers read 2608\.13546
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2608.13546 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2608.13546 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2608.13546 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
AlayaWorld: Interactive Long-Horizon World Modeling -- Full Technical Report
AlayaWorld is a 15B-parameter interactive video world model that generates 24-fps video at 540p and 720p, using autoregressive latent chunk generation with bounded visual context and distillation to reduce inference steps. It achieves state-of-the-art performance on the iWorld-Bench for long-horizon generation.
AlayaWorld: Long-Horizon and Playable Video World Generation
AlayaWorld is an open-source framework for building interactive generative worlds that enables real-time user interaction and supports diverse actions. It unifies the complete development pipeline from data preparation to deployment.
Baba in Wonderland: Online Self-Supervised Dynamics Discovery for Executable World Models
Introduces Alice, a closed-loop system that learns executable world models online under prior misalignment by treating failed candidate updates as structural signal, achieving improved performance on a variant of Baba Is You with semantically remapped labels.
Infinite Worlds with Versatile Interactions
This paper introduces LingBot-World 2.0, an advanced world modeling system with unbounded interaction horizons, real-time 720p 60fps video generation, diverse interactive elements (e.g., attacking, spell-casting), and novel multi-agent behavior control via pilot and director agents, along with a shared multi-player interface.
Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
Agent-World introduces a self-evolving training framework for general agent intelligence that autonomously discovers real-world environments and tasks via the Model Context Protocol, enabling continuous learning. Agent-World-8B and 14B models outperform strong proprietary models across 23 challenging agent benchmarks.