Autoregressive Video Generation

Overview

Autoregressive video generation studies how to produce a sequence of frames or video chunks conditioned on earlier content, enabling streaming output and interactive control. Common approaches use causal diffusion transformers, often distilled from bidirectional models to reduce denoising steps, and add action, camera, audio, or text conditioning for world models and editing. To extend generation beyond short clips, methods train on model-generated histories and use retrieval, compressed KV caches, or structured memory to preserve identities, scenes, and motion. The central challenge is balancing speed and bounded memory with visual quality, diverse motion, and consistency over long rollouts, which benchmarks such as VBench-Long and task-specific interaction tests attempt to measure.

Latest papers

226 papers26 papers in the last four weeks, down 38% on the four weeks before. 0.2% of all new papers.

Nov 10, 2026

BirdDreams: Joint Perspective and BEV Video Generation for a Plausible, Reactive Driving World Model

Driving world models aim to simulate how a scene unfolds under different ego actions. This requires plausible motion of surrounding vehicles as well as precise ego-trajectory control. Learning vehicle motion from perspective video alone is challenging because motion is entangled with appearance and vehicles can be occluded. Prior work therefore conditions video generation on layouts that specify vehicle motion, providing control over surrounding vehicles but requiring their future trajectories as inputs. We introduce BirdDreams, an autoregressive driving world model that jointly generates perspective video and a schematic bird's-eye-view (BEV) video, making scene layout part of the predicted future. The BEV, rendered from scene annotations for training, depicts roads and vehicles as simple shapes at a fixed metric scale, providing a top-down view of their positions and motion with reduced occlusion between vehicles. Perspective video, by contrast, retains scene details that the BEV omits, such as traffic signals and object appearance. Joint attention between two video diffusion transformers, one per stream, lets each stream inform the other during generation. The predicted BEV also provides a geometric interface for ego control: we translate and rotate its latest frame along an input ego trajectory and use the resulting frames as a soft condition that shows how the scene would appear to move if only the ego vehicle moved. At inference, BirdDreams generates both videos from a single camera image without requiring an HD map or vehicle-state inputs. On PhysicalAI-AV and nuScenes, BirdDreams follows input ego trajectories more faithfully than prior driving world models and generates more plausible ego motion without ego-action conditioning. Human evaluation shows that BirdDreams outperforms a perspective-only baseline in generating realistic surrounding-vehicle behavior.
Nov 8, 2026

Semantically Aligned KV Cache Compression for Efficient Agentic Autoregressive Video Generation

Agentic autoregressive video generation combines semantic planning, sequential generation, and visual verification for controllable synthesis. Growing visual histories enlarge the key-value (KV) cache, increasing memory use and attention costs across candidate rollouts. Existing compression methods primarily optimize individual generators without the planner's semantic constraints, risking low priority for states supporting requested attributes or actions. We introduce SemanticKV, a training-free method for semantically aligned KV cache compression. To our knowledge, SemanticKV is the first KV cache compression method for agentic autoregressive video generation that explicitly uses planner-extracted semantic constraints to guide retention. Text cross-attention identifies constraint-related video positions, whose self-attention queries are collected into separate groups. Attention weights are averaged within each group and maximized across groups, giving each cached entry a high score when strongly used by any represented constraint. A single top-k selection retains high-scoring entries under the memory budget, leaving their keys, values, and positions unchanged. For fixed query groups and equal entry costs, this selection minimizes an additive upper bound on worst-group discarded attention. The method operates independently within each candidate without retraining or additional video generations. We evaluate four AR video backbones on three benchmarks for 5-, 30-, and 60-second generation. On Causal Forcing at a compression ratio of 0.7, SemanticKV obtains a VBench Total Score of 84.10, compared with 84.05 for the strongest compressed baseline and 83.99 for Full KV. It achieves a 1.22x speedup over Full KV and reduces KV-cache memory from 6.04 GB to 1.81 GB, a 3.35x reduction.
Nov 6, 2026

AR-Edit: Training-Free Autoregressive Video Editing without Inversion

Editing a video stream in real time, without access to future frames, is essential for interactive applications. Existing training-free video editing methods, however, assume full offline access to the entire sequence, while streaming approaches either rely on additional training or are limited to specific editing techniques. We show that this limitation can be removed by analyzing the self-attention dynamics of DMD-distilled autoregressive (AR) video diffusion models. We uncover a key asymmetry: queries and keys remain stable and aligned with their source counterparts across denoising timesteps, while values diverge and encode the evolving content. Inherited from the non-distilled bidirectional teacher, this asymmetry is strongest at high noise, where few-step streaming models operate. The resulting mechanism is simple: a single forward pass suffices to capture source identity in the value features, which can be reused throughout generation. Starting from pure Gaussian noise and injecting only these source values reconstructs the input with fidelity matching or surpassing inversion-based methods, effectively eliminating inversion. Based on this insight, we introduce AR-Edit, a training-free, inversion-free method that caches source values once and selectively injects them during denoising. Across multiple AR backbones, AR-Edit enables real-time streaming editing with superior source preservation and tonal fidelity at minimal overhead.
Nov 6, 2026

Trajectory Forcing: Exploiting Diffusion Trajectories for Autoregressive Long Video Generation

Autoregressive (AR) modeling is arguably the only way to extend a fixed-length video diffusion model to longer or even streaming videos. However, existing AR diffusion methods use only the previously predicted chunk as context for the next, discarding the intermediate predictions that constitute the chunk's generation history. In contrast, conventional AR models such as LLMs condition on the entire history of preceding tokens. To this end, we propose Trajectory Forcing (TraF), an AR diffusion method that exploits the denoising trajectories of previous video chunks as context. TraF operates under two training regimes: (1) with ground-truth videos, where pseudo-trajectories are constructed from Gaussian interpolations and re-predicted by the model, and (2) with a teacher model, where real trajectories are obtained from the student's own denoising rollout. The first regime naturally provides a strong initialization for the second. On the 5-second VBench benchmark, TraF achieves a Total score of 85.15, outperforming Self Forcing (+0.84) and Causal Forcing (+1.11), with Quality and Semantic improving simultaneously. On 30-second generation---6×\times the training horizon---TraF achieves the highest overall quality among all compared methods.
Nov 5, 2026

Training-Free Predictive Spatio-Temporal Token Pruning for Few-Step Autoregressive Video Diffusion

Few-step autoregressive (AR) video diffusion models generate videos chunk by chunk and have made efficient video generation practical. However, further accelerating these models toward real-time generation remains challenging, as few-step distillation leaves little cross-step redundancy for existing training-free caching methods to exploit. We revisit this regime and identify two token-level redundancy patterns: cross-frame redundancy from static content shared across adjacent frames, and cross-step redundancy from tokens that converge early during denoising. Crucially, we find that early clean estimates along the model's own denoising trajectory provide an effective online signal for locating both patterns, requiring no retraining, auxiliary predictor, or reference video. Based on these findings, we propose PreSTP (Predictive Spatio-Temporal Token Pruning), a training-free framework that constructs a unified token-wise redundancy mask from cross-frame similarity and cross-step convergence. Tokens marked as redundant skip subsequent computation under a consistent last-computed-value restoration rule, and a content-adaptive mask reuse strategy further reduces dense probing cost. On 5-second generation, PreSTP achieves 1.52x-3.20x speedups and a better efficiency-fidelity trade-off than coarse-grained caching baselines such as TeaCache and FlowCache. On 30-second generation, it reaches up to 1.96x speedup while improving VBench-Long and suppressing long-horizon drift. Notably, it continues to reduce inference cost even in the extreme one-step regime, where cross-step redundancy vanishes entirely and coarse step-wise caching is no longer applicable.
Nov 4, 2026

NanoKV: High-Ratio KV Cache Compression for Autoregressive Video Generation

Autoregressive video models rely on historical key-value (KV) caches to preserve scene consistency over extended rollouts, making KV storage and repeated attention over history a growing memory and computational bottleneck. Aggressive compression can degrade generation quality, while reconstructing full-dimensional KV leaves attention computation largely unchanged. We introduce NanoKV, a frozen-backbone framework for generation-aware high-ratio KV compression. NanoKV exploits redundancy across tokens and attention heads by storing shared content in cluster centers and encoding joint head-channel residuals into low-bit coefficients. Crucially, NanoKV uses different fitting objectives for Keys and Values according to their distinct roles in attention: Key decoders are optimized to preserve downstream block outputs, whereas Value encoders and decoders are fitted for faithful reconstruction through the deployed quantization path. Building on the same compressed representation, the optional NanoKV-Fast variant reuses the stored coefficients for mixed-dimensional attention, reducing computation without modifying the video backbone. Across three autoregressive video backbones, NanoKV consistently improves the quality-storage trade-off. On HY-WorldPlay, NanoKV achieves 26.09×26.09\times effective KV compression, nearly twice that of KVTC, while improving PSNR by 1.0 dB. NanoKV-Fast retains 20.53×20.53\times effective compression and achieves an approximately 1.15×1.15\times end-to-end speedup over KVTC.
Nov 3, 2026

What Remains in Sight? Autoregressive Video Decoding as Representation-Guided Context Rewriting

Long-horizon video generation is increasingly moving toward autoregressive rollout, where each newly generated segment becomes part of the evidence used for future prediction. Recent causal decoders, long-context training, and KV-cache mechanisms have extended feasible duration, but most inference pipelines still decide the visible past implicitly through recent windows, compressed caches, or external memory. This raises a direct question: can the choice of which past frames or segments to show to the decoder be treated as an explicit test-time decision? We answer this question with ReCR, a training-free method that formulates autoregressive video decoding as representation-guided context rewriting. At each step, ReCR builds a selected visible context from the generated history by using internal decoder representations to favor context units that match the global rollout state, avoid redundancy with the recent continuation, preserve temporal boundary coverage, and widen the range of spatial layouts in view. The selected units are then mapped onto a compact causal axis before decoding the next block. Across long-horizon text-to-video benchmarks, ReCR consistently improves multiple autoregressive backbones under matched visible-token budgets. Empirical analyses such as representation-space diagnostics and transition-count prompt switching further show that improving what remains visible is an effective path toward more stable long-video decoding.
Nov 2, 2026

TaoMate-HD: Progressive-Resolution Autoregressive Audio-Video Generation

Streaming digital humans produce detailed video while continuing speech as new text arrives. High-resolution denoising is costly, and text updates can cause repeated or interrupted speech. We present TaoMate-HD, an autoregressive model for high-resolution audio-video streaming. For each short segment, or chunk, two denoising steps establish coarse video structure at low resolution. A final step refines the video to high resolution, with audio generated throughout the same trajectory. Each stage passes its audiovisual predictions to the same stage of the next chunk as history. Across text updates, the model retains this history to infer what has already been spoken. Continuation training utilizes asymmetric text conditioning, with separate text windows for generation and distillation. Experiments show a 1.6x speedup over direct high-resolution generation with the same model and hardware. Optimized stage-parallel deployment provides a further 1.9x overall speedup over serial progressive generation. Our continuation training reduces character error rate by 11.18 percentage points relative to the baseline. The model also supports hour-long audiovisual generation. Our project page is https://taoliveaigc.github.io/TaoMate-HD.
Nov 2, 2026

Disen-Forcing: Disentangling Semantic Anchoring from Motion for Autoregressive Video Diffusion

Autoregressive video diffusion models (AR-VDMs) enable real-time, long-horizon video generation but degrade rapidly once the inference horizon outruns the training horizon, due to compounding exposure bias. To stabilize long rollouts, recent methods heuristically adopt frame sink, keeping initial frames as a persistent model context, inspired by attention sinks in LLMs. We revisit this design and find that the mechanism by which initial frames stabilize long-range consistency in AR-VDMs is not an LLM-style attention sink; instead, long-range consistency is anchored through a sparse subset of attention heads, while the majority of heads predominantly capture local temporal dynamics. This observation reveals an over-conditioning pathology in existing frame-sink strategies: by universally exposing every head to the pinned initial-frame context, the model becomes prone to collapsing motion toward the low-dynamic modes of the teacher during model distillation. We propose Disen-Forcing, a distillation-time plug-in for AR-VDMs that disentangles and reinforces this head specialization through disentangled context routing and motion-disentangling regularization. On long-horizon video generation benchmarks, Disen-Forcing simultaneously improves long-range consistency and motion dynamics of base AR-VDMs, a trade-off that sink-based methods fail to balance.
Nov 2, 2026

History Forcing: Distilling History-Dependent Continuations for Long-Horizon Video World Models

Few-step autoregressive video diffusion enables efficient generation, yet maintaining consistency over long rollouts remains challenging, especially when subjects or scenes reappear after going out of sight. Despite progress in fidelity and rollout robustness, existing distillation methods do not fully exploit richer historical information to constrain the student's current prediction. To address this gap, we introduce History Forcing, a unified on-policy distillation framework that strengthens historical supervision for few-step autoregressive video generators. The framework uses the residual between teacher predictions with and without informative historical context to decouple this supervision. Such a historical context can be obtained using earlier generated chunks with less accumulated drift or a fixed-view teacher rollout that keeps the scene visible while the student looks away. The residual is then weighted and added to the score or velocity prediction used for distillation, forming a corrected target. By aligning the student with this corrected target through distillation, History Forcing incorporates long-range historical constraints into autoregressive generation. Our method achieves state-of-the-art performance among 1.3B models, with a VBench quality score of 85.07 for 15s video generation and a VBench-Long aggregate score of 79.82 for 60s generation. For out-of-sight world modeling, evaluations further demonstrate more plausible scene evolution and more consistent states when the camera returns, with Evolution Quality improving from 2.35 to 2.70 on PlayWorld and Object Consistency from 58.43 to 62.18 on MemoBench.
Nov 1, 2026

Learning-to-Memorize: Dynamic Context Management for Long-Horizon Autoregressive Video Generation

Autoregressive video generation has achieved impressive results on short clips, yet generating long videos remains challenging due to error accumulation over extended horizons, where each predicted frame depends on potentially imperfect previous frames. Existing approaches typically rely on static context management strategies, such as sliding-window caches that discard early frames or fixed sink frames that permanently preserve initial content. These designs either lose critical long-range context or anchor generation to outdated information, which can lead to degraded fidelity, motion stagnation, and reduced diversity. In this paper, we propose Learning to Memorize (L2M), a learned context memory management framework that dynamically retains, evicts, and evolves historical frames according to predicted importance. The framework consists of an importance prediction router that estimates the relevance of each historical frame, degradation-aware training that improves robustness to noisy or corrupted context, and a dynamic memory initialization and evolution mechanism that updates long-term memory using exponential moving average (EMA) importance scores. This allows high-quality frames to replace stale anchors and enables memory to adapt naturally to scene evolution. Trained with a teacher-student distillation objective and sparsity regularization, L2M makes efficient use of a fixed KV-cache budget while preserving the most informative contextual information. Extensive experiments demonstrate that L2M improves long-horizon video generation quality and temporal consistency over existing baselines, providing a learned strategy for context memory management in long autoregressive video generation.
Oct 31, 2026

What Helps Denoising Can Hurt Memory: Rethinking History in Long Video Generation

Autoregressive video generators produce long videos block by block, using historical context to maintain temporal consistency. After denoising, a committing pass encodes each clean block into keys and values (KV) that later blocks read, and the same clean block can be encoded into different KV depending on the history it reads. We therefore study historical context not only as a condition for denoising but also as an input to committing. Older historical context that supports denoising can harm subsequent generation when read during committing. At fixed context size and temporal positions, older historical context lowers mean clip-to-clip Subject Consistency. Across 5 generator configurations supporting longer contexts, consistency rises, plateaus, and then declines as committing context expands. Motivated by these findings, we propose Commit Forcing, a training-free rule that organizes committing into a chain. During committing, each block reads only the preceding block's stored KV, so older historical context reaches new entries only through the chain. We apply this rule to 6 pretrained generators. On 60-second VBench, Commit Forcing improves Quality and Total on all 6 generators, by up to 1.27 Quality points and 1.10 Total points. Long-range consistency gains extend to 240-second videos. In a blinded comparison on 120-second LongLive videos, Commit Forcing achieves win rates of 83.5% for subject and scene consistency and 82.9% for overall preference against the base generator, excluding ties.
Oct 30, 2026

Any Forcing: Any-Step Generality and One-Step Stability for Autoregressive Video Generation

Autoregressive video generation enables efficient inference through streaming prediction. However, existing methods are typically restricted to fixed sampling steps, limiting the flexibility to balance generation quality and efficiency, while one-step generation suffers from degraded visual quality and error accumulation over long sequences, leading to content drift and generation collapse. To address these issues, we propose Any Forcing, a unified framework for achieving Any-Step Generality and One-Step Stability. Specifically, we introduce a unified adversarial discriminator that jointly performs real-distribution alignment, inter-step consistency alignment, and first-frame quality enhancement, enabling flexible sampling while preserving one-step generation quality. Furthermore, we propose a training-free adaptive long-term memory selection mechanism that selects historical states based on global semantic relevance and local value contribution, while remapping their temporal positions to preserve temporal relationships within the memory cache, thereby enabling stable one-step long-video generation. Experiments demonstrate that Any Forcing achieves high-quality and stable one-step autoregressive video generation, achieving 20.9 FPS and 24.7 FPS for short and long video generation on a single H100 GPU, respectively.
Oct 27, 2026

Hierarchical Diffusion for Camera-controlled Autoregressive Video Generation

We address the task of long-video, camera-controlled generation from a single image. The dominant autoregressive paradigm suffers from two key limitations: (1) error accumulation over time, as distant frames are conditioned on previously generated outputs rather than ground truth, leading to drift; and (2) ambiguity in how to incorporate camera control. Standard solutions to both--such as self-forcing and conditioning on explicit 6-DOF poses--either incur significant computational overhead or produce drifting scene geometries over time. We propose a simple and lightweight framework that combines autoregressive generation (WanAR) with hierarchical long-context denoising (FixAR). To enforce scene geometry remains consistent with input cameras, we draw from feed-forward novel-view synthesis and reformulate video generation as a warping-and-inpainting problem. To mitigate drift, we introduce a hierarchical denoising strategy in which different stages condition on different temporal histories. Specifically, we first generate frames conditioned on the immediate past, and then refine them using a second denoiser conditioned on a temporally subsampled long-range history to enforce global consistency. This fine-to-coarse scheme enables each generated frame to effectively condition on the full temporal context without incurring a prohibitive cost. Experiments on RealEstate10k and DL3DV demonstrate that WanAR+FixAR outperforms prior methods for camera-controlled long-video generation, producing results that are both temporally stable and consistent with input camera trajectories.
Oct 26, 2026

BoxWorld: Let World Engines Control and Video Generators Render

We tackle the challenge of building real-time interactive video worlds with user-defined behaviors. Such worlds must track object states, apply interaction rules, and render the resulting consequences as users act. Yet existing end-to-end video world models typically learn interaction dynamics within the generator, making it difficult to specify object-level rules and reason about their consequences. In this work, we introduce BoxWorld, which separates interaction reasoning from generative rendering. Given user-specified rules, a coding agent creates a Box World Engine that determines interaction outcomes and exposes the resulting object states through a lightweight interface of layout boxes, persistent object attributes, and dynamic status variables, leaving appearance and physical effects to generative rendering. A Box-State Generative Renderer then renders this evolving state stream into video autoregressively, enabling online control over object motion, state switches, and conditional events. Rendering this state stream is non-trivial: the renderer must respond to object-level state updates in real time while remaining coherent with the generated history. To this end, we introduce a three-stage training strategy that learns object grounding from open-domain videos and state transitions from gameplay videos, both annotated automatically, and finally performs rollout distillation on self-generated histories. We further propose PairedStateBench, a paired benchmark for fine-grained interaction control. BoxWorld substantially outperforms end-to-end world models and remains competitive across multiple public benchmarks. Project page: https://boxworld-web.github.io.
Oct 25, 2026

Learning the Autoregressive: Direct Score-Difference Estimation with Amortized Future Credit for Video Distillation

Few-step distillation for autoregressive (AR) video generation commonly first initializes an AR student and then refines it with distribution matching distillation (DMD). This refinement process faces two challenges. First, although initialization already yields a reasonably capable student, DMD obtains the refinement direction indirectly by estimating the full student score before forming the teacher-student correction. Second, detached-history training prevents the generator from assigning credit to a current chunk for its influence on subsequent chunks. We introduce Leta, a few-step distillation framework that addresses these challenges with ScoreDelta and FutureCredit. ScoreDelta directly estimates the teacher-student correction from denoising residuals, while FutureCredit learns an auxiliary cross-chunk update signal from vector-Jacobian products (VJPs) of next-chunk teacher residuals. Leta combines these signals into a joint teacher-guided update field in a single forward pass during generator updates. Teacher evaluations and credit VJPs are used only for estimator training. For ScoreDelta, we establish population equivalence to score-difference estimation and a conditional approximation advantage under limited capacity in a fixed-feature linear model. In our controlled experiments, Leta improves over DMD by 0.9 points on VBench and 0.66 on VideoAlign. This advantage persists at a higher generator update frequency (a generator-to-estimator update ratio of G:E=1:1), where Leta improves over DMD by 1.6 on VBench and 0.35 on VideoAlign. Project page: https://scdzx.github.io/Leta/.
Oct 24, 2026

Mix Forcing: Causal Pre-Training via Mixture Attention for Interactive World Models

While interactive video world models require autoregressive generation, directly adapting pretrained bidirectional video diffusion models into autoregressive ones with teacher forcing can disrupt their visual priors and degrade autoregressive generation quality. In this paper, we introduce Mix Forcing, a causal pre-training paradigm for interactive video world models, implemented with the Mixture-of-Bidirectional-and-Autoregressive (MoBA) attention. Mix Forcing jointly trains a video diffusion model with both teacher forcing and bidirectional training, yielding a strong causal initialization for few-step post-training. In self-attention, Mix Forcing leverages teacher forcing to learn causal next-chunk prediction for autoregressive rollout, while using bidirectional attention to preserve pretrained visual priors. To support online text interaction, Mix Forcing further adopts a chunk-wise cross-attention design that conditions each video chunk on a shared environment prompt and its own local prompt, without exposing future instructions to the current chunk. Following causal pre-training on a 14B image-to-video model, we apply consistency and distribution matching distillation to obtain a four-step autoregressive generator. Experiments on our internal benchmark and public benchmarks, including WBench and SANA-WM-Bench, show that our model achieves state-of-the-art visual quality and timely event following with precise camera control. Notably, our model supports high-fidelity long-horizon generation without quality degradation even over one-hour rollouts. Project page is here: https://zelingao98.github.io/mixforcing
Oct 22, 2026

Direct Forcing: Single-Stage Training with Representation-Space Supervision for Few-Step Autoregressive Video Generation

Few-step autoregressive video generation enables low-latency streaming, yet prevailing training pipelines separate causal adaptation from distribution matching. Recent representation-based text-to-video methods retain this separation by starting with an ODE initialization stage, even when the target distribution is defined directly by reference videos. We introduce \method, the first single-stage framework that uses reference videos both as paired examples of how a history continues and as a target population for self-generated rollouts. Specifically, Direct Forcing is trained with two complementary signals: (1) Causal Continuation Alignment learns paired spatiotemporal correspondence under reference histories; and (2) Rollout Distribution Matching employs maximum mean discrepancy to align the representation distributions of self-generated and reference videos. To coordinate these signals, we introduce Dynamic Gradient Balancing, which adapts their relative gradient contributions during training. Experiments with Wan2.1-1.3B show that Direct Forcing learns a four-step autoregressive generator in 250 updates without a separate ODE initialization stage or online diffusion-score supervision, achieving 84.63 VBench Total and 86.01 Quality.
Oct 21, 2026

Looking Ahead: Modeling Future Evolution in Autoregressive Video Generation

Autoregressive video diffusion models generate videos chunk by chunk, yet producing plausible continuations does not guarantee that they capture how scenes evolve. Under teacher forcing, strong temporal redundancy may allow local visual continuity to serve as a shortcut for current-chunk denoising, reducing the pressure to model visual evolution. We introduce Future Chunk Prediction (FCP), an auxiliary objective jointly optimized with teacher-forced denoising that predicts temporally separated chunks without access to the intervening visual context. Drawing supervision directly from training videos, FCP requires no additional pretrained visual encoder and leaves autoregressive inference unchanged. Our method achieves superior physical plausibility over standard teacher-forced autoregressive training with a lower overall training compute budget, highlighting the effectiveness and efficiency of future-chunk supervision for modeling visual evolution. Intriguingly, our visualizations suggest that FCP fosters a more structured representation space, which may be valuable for building autoregressive world models.
Oct 21, 2026

LiveAct: Towards Real-Time Multi-Agent World Models with Heterogeneous Action Interfaces

Interactive video world models predict how an environment evolves under user actions, but most existing systems assume one controlled agent and one task-specific action interface. We study multi-agent video world modeling, where one model generates a joint visual future for multiple controlled agents while interpreting heterogeneous controls ranging from binary game buttons to continuous robot commands. This task presents two complementary challenges. First, every agent-indexed action stream must control its intended visual entity independently. Second, the model must adapt the meaning of each control variable across different tasks, where the same tensor dimension may denote an attack, steering command, or robot command. We introduce LiveAct, centered on a Heterogeneous Action Interface (HAI) whose two components mirror this decomposition. To support independent control, Agent-Aware Spatial Modulation uses language-driven spatial gates and agent-specific attention to bind each action stream to its intended visual agent. To accommodate heterogeneous controls, Schema-Guided Action Adaptation describes each control variable in language and modulates its semantic token with the synchronized numerical value. We further transfer the complete interface to a blockwise causal generator through teacher forcing and rollout-aware distribution matching, enabling efficient streaming generation while retaining agent-specific control and action semantics. We instantiate the formulation on six multi-agent sources spanning games and multi-robot manipulation, and evaluate visual quality, action following, agent attribution, and latency. Across games and robot manipulation, LiveAct reaches 0.898 and 0.942 agent-attribution accuracy without agent locations or identity crops, and the causal generator attains 17.70 FPS at 512×448512\times 448.
Oct 20, 2026

Offline Makes Online Better: Calibrated Cache Compression for Video Generation

Token merging is an effective approach to reducing KV-cache memory consumption in autoregressive video generation by clustering temporally redundant tokens and retaining only a small set of representatives. Existing methods, however, typically apply a uniform merging ratio across all attention layers and heads. We observe that different layers and heads exhibit substantially different yet jointly correlated sensitivities to token merging, and that these sensitivity patterns remain consistent across prompts. Based on this observation, we introduce CALM, a Calibrated Allocation for Layer-Head Merging that uses offline calibration to improve online cache compression. Exhaustively evaluating all possible layer-head allocations is prohibitively expensive. To enable efficient calibration, we show that attention-output error strongly correlates with end-to-end video quality and can therefore serve as a lightweight proxy for allocation risk assessment. We evaluate different allocation strategies using this proxy and construct a fixed layer-head risk map offline. Given a cache budget at inference time, CALM uses dynamic programming to determine the best allocation of merging ratios. Experiments across three different autoregressive video models demonstrate that CALM consistently outperforms uniform token merging under the same memory budget, by reducing the LPIPS to original videos by around 21% with higher video-quality scores.
Oct 20, 2026

Matrix-Game 4.0: Harnessing Video Models for Real-Time Playable Game Worlds

Moving from controllable videos to genuinely playable worlds requires persistent world state, executable game rules, and low-latency interaction. We outline a roadmap for the next generation of Matrix-Game that separates game execution from visual generation through an executable game harness, and introduce Matrix-Game 4.0 as the first system built on this design. A lightweight game engine maintains world state and executes interaction rules, a multimodal coding agent translates player ideas into game code and dynamic scene captions, and a renderer based on autoregressive video model transforms engine-rendered low-poly proxies into detailed visuals, conditioned on dynamic captions and memory references. This framework provides continuous guidance for visual generation and enables natural-language creation and modification of scenes and gameplay. Our renderer is trained on aligned low-poly and high-fidelity video pairs and distilled into a causal generator that generates in four-frame chunks with one or two denoising steps. Together with targeted inference optimizations, Matrix-Game 4.0 enables 480p, 60 FPS gameplay on a single NVIDIA RTX5090 with a key-press-to-display latency of 112ms. Single-step generation achieves 83FPS steady-state throughput with 18.2GiB of peak GPU memory usage. With fixed prompts, steady-state throughput also reaches 67FPS on an RTX5080 and 58FPS on an RTX5070Ti. We demonstrate customizable shooter and driving games with engine-enforced mechanics, and release code, model weights, and low-poly game assets to support research and creation of playable game worlds.
Oct 20, 2026

Recast the Past: Stable Long Video Generation with Full Context Memory

Long-horizon video generation requires retaining an ever-expanding history, but conditioning on its growing token sequence is costly. Existing methods restrict or compress historical context, potentially losing information needed for future generation. This work proposes ReCam, which groups local spatiotemporal blocks of historical latents into vectors and projects them directly into the high-dimensional hidden space of a Diffusion Transformer. The model learns to use the resulting memory tokens through native self-attention, reducing token count while retaining block-level access to the full history. To mitigate temporal drift during autoregressive rollout, we further introduce dual-pointer denoising with separate noise schedules for recent context and target frames. Among the evaluated methods, ReCam achieves the highest VBench-Long overall score of 80.1 and competitive static-scene revisitation performance on R2M-Bench. It also leads on all four consistency metrics for off-screen subject reappearance, including clothing consistency of 4.09 versus 3.03 for the best baseline.
Oct 19, 2026

Dynamics-Aware Sparse Attention for Efficient Autoregressive Video Diffusion

Autoregressive Diffusion Transformers have emerged as a powerful paradigm for long video generation. However, they are often constrained by prohibitive computational overhead and substantial memory footprints. This inefficiency arises from dense computation mechanisms that allocate uniform processing resources across all spatiotemporal regions, failing to leverage the inherent temporal redundancy of video where motion is typically confined to sparse areas. In this paper, we propose Dynamics-Aware Sparse Attention (DASA), a unified framework that shifts from dense computation to sparse, motion-driven generation to eliminate both storage and computational redundancies. Inspired by video codec principles, our approach explicitly decouples video content into dynamic and static regions. To mitigate memory bottlenecks, we selectively compress the historical KV cache by retaining only dynamic features within the generated chunk. Furthermore, to reduce computational overhead, we employ a region-level computation allocation strategy based on localized dynamics. Additionally, a dynamics-guided distillation method is introduced to enhance performance via a lightweight distillation training process. Experiments on VBench demonstrate that our approach accelerates autoregressive video generation by 2.16x, while preserving high visual fidelity.
Oct 19, 2026

TriTD: Tri-Partite Trajectory-Distribution Distillation for Real-Time Autoregressive Video Generation

Autoregressive (AR) video diffusion has emerged as a leading paradigm for efficient video generation, yet its practical deployment remains bottlenecked by the trade-off between generation quality and inference efficiency. We attribute this bottleneck to three error sources in AR rollouts: the base model error, the train-inference distribution mismatch, and the upstream error propagation. Building on this decomposition, we derive closed-form upper bounds on the per-step and cumulative errors, providing a principled basis for jointly optimizing generation quality and inference efficiency. Guided by this analysis, we propose TriTD, a Tri-Partite Trajectory-Distribution co-optimized distillation framework with three complementary modules. First, for ODE initialization, we adopt a Next-Node Trajectory Distillation (NTD) objective, which shortens the supervision horizon and admits a lower regression-error floor. Second, we introduce Chunk-Graded Noising (CGN), a monotonically non-decreasing per-chunk noising schedule that helps narrow the expected train-inference noise mismatch and removes redundant KV-cache updates, yielding substantial speedups at no quality cost. Third, we propose Orthogonal-Parallel DMD (OP-DMD), a distribution-matching loss that retains the original DMD supervision while introducing an additional regression constraint in the orthogonal subspace to suppress off-axis drift, improving both training stability and final generation quality. Extensive experiments show that TriTD consistently surpasses existing baselines in visual quality while reaching a real-time throughput of 26.8 FPS on a single H100 GPU, delivering a balanced quality-efficiency solution for interactive streaming video generation.
Oct 19, 2026

Clean Forcing: Drift-Resistant Autoregressive Video Diffusion with a Frozen Base

Autoregressive (AR) video diffusion models generate streaming video of arbitrary length by conditioning each new chunk of frames on those already generated. However, they suffer from exposure bias, where a model trained on clean context must continue from its own imperfect outputs, and errors compound until quality collapses within seconds (drift). Existing methods either require cluster-scale base retraining, apply training-free inference corrections with inconsistent results, or train frozen-base correctors for sampling and cache errors. We instead find that drift is largely deterministic and therefore learnable, as holding the noisy state fixed while varying only the history reveals a counterfactual velocity gap that is ~95% systematic across noise seeds. Based on this observation, we introduce Clean Forcing, which trains a 5.9M-parameter LoRA corrector on the frozen base model to regress this gap, with a closed-loop stage that adds DAgger-style aggregation and a drift-contraction objective to expose the corrector to its own rollouts. Because the clean histories can be the base model's own single-shot generations, training needs no real videos, and the merged corrector adds no inference cost. In 50 s text-to-video generation, Clean Forcing reduces Δ-drift from +11.3 to +1.65 without real video and to -0.16 with 40 real clips, improving on the best published result of +3.57 with over 150x less data. Extensive experiments demonstrate that Clean Forcing outperforms prior methods in aesthetic quality and data efficiency with reduced drift for long-horizon video generation.
Oct 15, 2026

Speculative Forcing for Efficient Autoregressive Video Generation

Few-step autoregressive video generators enable streaming generation, yet they run every token through the final denoising step, regardless of how much refinement it needs. We find this uniform computation is often unnecessary: many tokens already closely match their eventual full-step outputs in the initial steps, while others require refinement until the end. We introduce Speculative Forcing, a framework that treats intermediate clean estimates as drafts and predicts when individual tokens can be safely accepted early. Once accepted, a token is frozen as its final output and undergoes no further denoising, while its cached keys and values remain available as context for the remaining tokens. We explore two alternative verifier designs: a training-free one based on similarity to clean content and stability across steps, and a learned one that reads the frozen generator's intermediate features to estimate each token's remaining gap to its full-step output. Across short-video, long-video, and world generation, Speculative Forcing preserves the quality of full-step sampling while substantially reducing computation. On LongLive, it matches the four-step generator with ~1.7 denoising steps per token, running over 1.5x faster at 41.7 and 37.8 fps. On Lingbot-World 2.0, where camera motion continually changes the visible content, it maintains quality with ~2.5 denoising steps per token and runs 1.35x faster at 24.5 fps.
Oct 14, 2026

Speculative Sampling in Interactive Video Generation

Interactive video generation commonly relies on small bidirectional chunks to reduce response latency. Yet our preliminary experiments show that larger chunks improve video quality and better utilize GPU parallelism, allowing more latents to be sampled with only a modest increase in sampling time. Motivated by these observations, we propose Speculative Sampling, a strategy that combines the modeling power of large bidirectional windows with the low-latency interaction of small ones. It expands the modeling window by appending noise latents as padding and caches the resulting predictions for reuse when the condition remains unchanged. Predictions are delivered incrementally, while the uncommitted future remains replaceable. Our latency analysis shows that this removes the waiting-time lower bound imposed by chunk boundaries, shifting the optimization focus from reducing chunk size to accelerating per-chunk sampling. Bidirectional window size thus becomes a decision variable for jointly optimizing visual quality and sampling time, rather than a necessary constraint on low-latency interaction. This opens a path to serving large bidirectional models in interactive streaming systems without complex streaming post-training.
Oct 14, 2026

Glance Forcing: Making Bidirectional Diffusion Autoregressive at a Glance

Autoregressive generation underlies real-time, causal video synthesis and is the backbone of video world models. Today's dominant approach distills a bidirectional diffusion model into a causal few-step student, but existing pipelines make this conversion prohibitively expensive, requiring offline Probability Flow ODE} (PF-ODE) trajectories that cost tens of thousands of GPU-hours and terabytes of storage. We show it can be made dramatically more efficient. Pushing this efficiency to its extreme, we reduce trajectory supervision to a single sample, the strictest setting possible, and find that it causes severe quality collapse. We trace this to the teacher's trajectory geometry: the high-noise regime, governing semantic layout and motion, and the low-noise regime, governing fine texture exhibit fundamentally different curvature and optimization behavior, so a single monolithic student cannot jointly fit both, causing severe representation interference. Motivated by this, we propose Glance Forcing, a lightweight distillation framework that decouples the denoising trajectory into two complementary slow and fast experts, each guided by a tailored, stage-specific objective, while keeping the pretrained teacher entirely frozen. This decoupling further reveals a general mechanism underlying few-step distillation: PF-ODE distillation mainly straightens the few-step trajectory geometry, while the subsequent distribution-matching stage independently governs distributional quality. Using only one PF-ODE trajectory, Glance Forcing already surpasses the fully-trained state-of-the-art baseline, raising the VBench Total score from 84.04 to 84.22 while cutting trajectory-collection cost from thousands of GPU-hours to about one A100-GPU-hour, with markedly larger gains in dynamic degree, VisionReward, and instruction-following, and consistent preference from human raters over strong baselines.
Oct 14, 2026

RepForcing: Representation-Space Initialization for Autoregressive Video Distillation

Distilling pretrained video diffusion models into few-step autoregressive (AR) generators typically combines student initialization with distribution matching distillation (DMD). An intermediate AR teacher can provide causally compatible targets for ODE initialization. However, causally compatible targets alone do not ensure high-quality generation under a short sampling schedule. We analyze the posterior-mean predictions favored by direct latent regression and empirically find that causal ODE initialization alleviates blur but still leaves fine detail unresolved. To address this limitation, we propose Representation Forcing, which learns few-step causal initialization through representation-space supervision rather than intermediate-teacher trajectory regression. The student is trained with teacher forcing by comparing clean predictions and training targets through normalized features of a frozen video diffusion transformer. This phase reuses the synthetic videos generated by a bidirectional model for AR teacher training and replaces both teacher training and causal ODE distillation, without generating additional causal ODE trajectories. The subsequent asymmetric DMD procedure remains unchanged. Empirical results show that, at four sampling steps, RepForcing better preserves generation quality than intermediate-teacher-based initialization before DMD. After DMD, our four-step chunk-wise model improves VBench Total from 84.04 to 84.66 (+0.62 points) and VisionReward from 6.326 to 10.074 (+59.25%) over the reported causal-ODE-based baseline.