# Fast Autoregressive Video Diffusion and World Models with Temporal Cache Compression and Sparse Attention

arXiv: https://arxiv.org/abs/2602.01801 · HTML: https://mlsanity.com/paper/2602.01801/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- Placed in Video-generation world models and simulators, one of 11 directions in World models.
- Builds on 17 of its 21 references that are in the map.
- First posted on arXiv 2 Feb 2026.

## Placed in

- World models / [Video-generation world models and simulators](https://mlsanity.com/wm/video-generation-world-models-and-simulators/)
- Diffusion models / [Video generation](https://mlsanity.com/diffusion/video-generation/)
- Efficient LLM inference and serving / [Fast autoregressive visual generation](https://mlsanity.com/inference/fast-autoregressive-visual-generation/)

## Builds on (most cited first, 17 of 17)

- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [Reformer: The Efficient Transformer](https://mlsanity.com/paper/2001.04451/) (2020)
- [HunyuanVideo: A Systematic Framework For Large Video Generative Models](https://mlsanity.com/paper/2412.03603/) (2024)
- [Cosmos World Foundation Model Platform for Physical AI](https://mlsanity.com/paper/2501.03575/) (2025)
- [Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphs](https://mlsanity.com/paper/1603.09320/) (2016)
- [Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion](https://mlsanity.com/paper/2506.08009/) (2025)
- [LTX-Video: Realtime Video Latent Diffusion](https://mlsanity.com/paper/2501.00103/) (2024)
- [From Slow Bidirectional to Fast Autoregressive Video Diffusion Models](https://mlsanity.com/paper/2412.07772/) (2024)
- [FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving](https://mlsanity.com/paper/2501.01005/) (2025)
- [Rolling Forcing: Autoregressive Long Video Diffusion in Real Time](https://mlsanity.com/paper/2509.25161/) (2025)
- [Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity](https://mlsanity.com/paper/2502.01776/) (2025)
- [Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation](https://mlsanity.com/paper/2505.18875/) (2025)
- [HunyuanWorld 1.0: Generating Immersive, Explorable, and Interactive 3D Worlds from Words or Pixels](https://mlsanity.com/paper/2507.21809/) (2025)
- [Radial Attention: O(nlog n) Sparse Attention with Energy Decay for Long Video Generation](https://mlsanity.com/paper/2506.19852/) (2025)
- [Emergent Temporal Correspondences from Video Diffusion Transformers](https://mlsanity.com/paper/2506.17220/) (2025)
- [Story2Board: A Training-Free Approach for Expressive Storyboard Generation](https://mlsanity.com/paper/2508.09983/) (2025)
- [OmnimatteZero: Fast Training-free Omnimatte with Pre-trained Video Diffusion Models](https://mlsanity.com/paper/2503.18033/) (2025)

## Built on by (most cited first, 18 of 18)

- [AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories](https://mlsanity.com/paper/2602.14941/) (2026)
- [Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Models](https://mlsanity.com/paper/2605.09681/) (2026)
- [Addressable Memory for Video World Models](https://mlsanity.com/paper/2608.07408/) (2026)
- [Echo-Forcing: A Scene Memory Framework for Interactive Long Video Generation](https://mlsanity.com/paper/2605.16003/) (2026)
- [Future Forcing: Future-aware Training-free KV Cache Policy for Autoregressive Video Generation](https://mlsanity.com/paper/2605.30083/) (2026)
- [TetherCache: Stabilizing Autoregressive Long-Form Video Generation with Gated Recall and Trusted Alignment](https://mlsanity.com/paper/2606.13035/) (2026)
- [Attend Locally, Remember Linearly: Linear Attention as Cross-Frame Memory for Autoregressive Video Diffusion](https://mlsanity.com/paper/2605.16579/) (2026)
- [Quantized Keys Steal Attention: Bias Correction for KV-Cache Compression in Video Diffusion](https://mlsanity.com/paper/2605.26266/) (2026)
- [Partition the Support, Reconstruct the Residual: Training-Free Sparse Attention for Video Generation and World Models](https://mlsanity.com/paper/2608.18484/) (2026)
- [SparSTAR: Sparse Attention for SpaceTime AutoRegressive Video Synthesis](https://mlsanity.com/paper/2608.10519/) (2026)
- [From Generation to Simulation: How Far Are World Models from Being True Simulators?](https://mlsanity.com/paper/2608.23070/) (2026)
- [In-Flight KV Cache with Clean Anchors for Faster Autoregressive Video Diffusion](https://mlsanity.com/paper/2609.32540/) (2026)
- [DeCoPrune: Efficient KV-Cache Pruning for Autoregressive Video Diffusion via Denoising Consistency](https://mlsanity.com/paper/2609.39096/) (2026)
- [PixReenact: Pixel-Conditioned Causal Video Diffusion for Streaming Head-Avatar Reenactment](https://mlsanity.com/paper/2610.05233/) (2026)
- [HLA-WM: Hybrid Linear Attention for Long-Horizon Video World Models](https://mlsanity.com/paper/2610.05739/) (2026)
- [CORAL-KV: Compact Cross-Layer Prediction and Residual Coding for Autoregressive Video KV Caches](https://mlsanity.com/paper/2610.12514/) (2026)
- [FrameFold: Response-Guided KV Cache Consolidation for Autoregressive Long Video Generation](https://mlsanity.com/paper/2610.14568/) (2026)
- [Into the LidarVerse: A Real-Time Unified LiDAR World Model Across Tasks and Sensors](https://mlsanity.com/paper/2610.19201/) (2026)

## Abstract (arXiv)

Autoregressive video diffusion models enable streaming generation, opening the door to long-form synthesis, video world models, and interactive neural game engines. However, their core attention layers become a major bottleneck at inference time: as generation progresses, the KV cache grows, causing both increasing latency and escalating GPU memory, which in turn restricts usable temporal context and harms long-range consistency. In this work, we study redundancy in autoregressive video diffusion and identify three persistent sources: near-duplicate cached keys across frames, slowly evolving (largely semantic) queries/keys that make many attention computations redundant, and cross-attention over long prompts where only a small subset of tokens matters per frame. Building on these observations, we propose a unified, training-free attention framework (FAST-AR) for FAST-AutoRegressive diffusion, consisting of three components: TempCache compresses the KV cache via temporal correspondence to bound cache growth; AnnCA accelerates cross-attention by selecting frame-relevant prompt tokens using fast approximate nearest neighbor (ANN) matching; and AnnSA sparsifies self-attention by restricting each query to semantically matched keys, also using a lightweight ANN. Together, these modules reduce attention, compute, and memory and are compatible with existing autoregressive diffusion backbones and world models. Experiments demonstrate up to x5 - x10 end-to-end speedups while preserving near-identical visual quality and, crucially, maintaining stable throughput and nearly constant peak GPU memory usage over long rollouts, where prior methods progressively slow down and suffer from increasing memory usage.