mlsanityData to 14 Oct 2026

Flash-VStream: Efficient Real-Time Understanding for Long Video Streams

arXiv 2506.23825 ↗First posted 30 Jun 2025cs.CVcited by 81

On the map: Multimodal LLMs / vision-language models → Video understanding with MLLMs

Abstract (arXiv)

Benefiting from the advances in large language models and cross-modal alignment, existing multimodal large language models have achieved prominent performance in image and short video understanding. However, the understanding of long videos is still challenging, as their long-context nature results in significant computational and memory overhead. Most existing work treats long videos in the same way as short videos, which is inefficient for real-world applications and hard to generalize to even longer videos. To address these issues, we propose Flash-VStream, an efficient video language model capable of processing extremely long videos and responding to user queries in real time. Particularly, we design a Flash Memory module, containing a low-capacity context memory to aggregate long-context temporal information and model the distribution of information density, and a high-capacity augmentation memory to retrieve detailed spatial information based on this distribution. Compared to existing models, Flash-VStream achieves significant reductions in inference latency. Extensive experiments on long video benchmarks and comprehensive video benchmarks, i.e., EgoSchema, MLVU, LVBench, MVBench and Video-MME, demonstrate the state-of-the-art performance and outstanding efficiency of our method. Code is available at https://github.com/IVGSZ/Flash-VStream.

Key facts. Cited by 81 papers in the map (data through 14 Oct 2026). Placed in Video understanding with MLLMs, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 78 of its 91 references that are in the map. First posted on arXiv 30 Jun 2025.

Builds on

78 of 91 references.

Closest ancestors · 8

Same direction, newest first

  1. 2024Kangaroo: A Powerful Video-Language Model Supporting Long-context Video Input28 Aug
  2. 2024LongVILA: Scaling Long-Context Visual Language Models for Long Videos19 Aug
  3. 2024Long Context Transfer from Language to Vision24 Jun
  4. 2024LVBench: An Extreme Long Video Understanding Benchmark12 Jun
  5. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis31 May
  6. 2023TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding4 Dec
  7. 2023LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models28 Nov
  8. 2023MVBench: A Comprehensive Multi-modal Video Understanding Benchmark28 Nov

Other references · 70

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2023GPT-4 Technical Reportcited by 18,396
  4. 2022Training language models to follow instructions with human feedbackcited by 15,971
  5. 2024The Llama 3 Herd of Modelscited by 14,715
  6. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  7. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  8. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  9. 2023Visual Instruction Tuningcited by 7,869
  10. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  11. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  12. 2023Improved Baselines with Visual Instruction Tuningcited by 4,418
38 more
  1. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  2. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  3. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  4. 2023InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuningcited by 2,876
  5. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,719
  6. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
  7. 2024Qwen2 Technical Reportcited by 2,120
  8. 2023Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Modelscited by 1,062
  9. 2021CLIP-Adapter: Better Vision-Language Models with Feature Adapterscited by 1,026
  10. 2024VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMscited by 771
  11. 2021NExT-QA:Next Phase of Question-Answering to Explaining Temporal Actionscited by 744
  12. 2023EgoSchema: A Diagnostic Benchmark for Very Long-form Video Language Understandingcited by 680
  13. 2019ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answeringcited by 561
  14. 2023MovieChat: From Dense Token to Sparse Memory for Long Video Understandingcited by 542
  15. 2022XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Modelcited by 439
  16. 2021Dynamic Neural Networks: A Surveycited by 408
  17. 2024ShareGPT4Video: Improving Video Understanding and Generation with Better Captionscited by 388
  18. 2025Kimi-VL Technical Reportcited by 388
  19. 2023Chat-UniVi: Unified Visual Representation Empowers Large Language Models with Image and Video Understandingcited by 371
  20. 2019Towards Real-Time Multi-Object Trackingcited by 311
  21. 2024VideoLLM-online: Online Video Large Language Model for Streaming Videocited by 225
  22. 2020MovieNet: A Holistic Dataset for Movie Understandingcited by 206
  23. 2023Self-Chained Image-Language Model for Video Localization and Question Answeringcited by 203
  24. 2024Streaming Long Video Understanding with Large Language Modelscited by 167
  25. 2024Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolutioncited by 159
  26. 2019SCSampler: Sampling Salient Clips from Video for Efficient Action Recognitioncited by 134
  27. 2023RoboVQA: Multimodal Long-Horizon Reasoning for Roboticscited by 120
  28. 2024ATP-LLaVA: Adaptive Token Pruning for Large Vision Language Modelscited by 89
  29. 2024Streaming Dense Video Captioningcited by 82
  30. 2024VoCo-LLaMA: Towards Vision Compression with Large Language Modelscited by 79
  31. 2021Adaptive Focus for Efficient Video Recognitioncited by 77
  32. 2023Vista-LLaMA: Reducing Hallucination in Video Language Models via Equal Distance to Visual Tokenscited by 71
  33. 2018A Twofold Siamese Network for Real-Time Object Trackingcited by 62
  34. 2021FrameExit: Conditional Early Exiting for Efficient Video Recognitioncited by 59
  35. 2023SOC: Semantic-Assisted Object Cluster for Referring Video Object Segmentationcited by 57
  36. 2022MIST: Multi-modal Iterative Spatial-Temporal Transformer for Long-form Video Question Answeringcited by 55
  37. 2024VideoLLM-MoD: Efficient Video-Language Streaming with Mixture-of-Depths Vision Computationcited by 55
  38. 2016Real-time Action Recognition with Enhanced Motion Vector CNNscited by 51

50 most cited of 70 All 70 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 10; 2026: 71.

81 papers.

Sort:

By area:

  1. 2025Thinking With Videos: Multimodal Tool-Augmented Reinforcement Learning for Long Video Reasoningcited by 91
  2. 2025StreamingTOM: Streaming Token Compression for Efficient Video Understandingcited by 43
  3. 2026A Simple Baseline for Streaming Video Understandingcited by 33
  4. 2026Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneouslycited by 22
  5. 2026OASIS: On-Demand Hierarchical Event Memory for Streaming Video Reasoningcited by 19
  6. 2026Thinking in Streaming Videocited by 16
  7. 2025A Unified Reasoning Framework for Holistic Zero-Shot Video Anomaly Analysiscited by 12
  8. 2025Thinking With Bounding Boxes: Enhancing Spatio-Temporal Video Grounding via Reinforcement Fine-Tuningcited by 12
  9. 2026EventMemAgent: Hierarchical Event-Centric Memory for Online Video Understanding with Adaptive Tool Usecited by 12
  10. 2025ScoreHOI: Physically Plausible Reconstruction of Human-Object Interaction via Score-Guided Diffusioncited by 9
  11. 2026Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospectscited by 9
  12. 2025VG-Refiner: Towards Tool-Refined Referring Grounded Reasoning via Agentic Reinforcement Learningcited by 8
38 more
  1. 2026Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Modelscited by 8
  2. 2026Ego-Grounding for Personalized Question-Answering in Egocentric Videoscited by 7
  3. 2026What Should a Streaming Video Model Remember?cited by 7
  4. 2026Think-as-You-See: Streaming Chain-of-Thought Reasoning for Large Vision-Language Modelscited by 6
  5. 2026SVCBench: A Streaming Video Counting Benchmark for Spatial-Temporal State Maintenancecited by 6
  6. 2026OVO-S-Bench: A Hierarchical Benchmark for Streaming Spatial Intelligence in Multimodal LLMscited by 6
  7. 2026EgoSAT: A Comprehensive Benchmark of Egocentric Streaming Interaction Understandingcited by 6
  8. 2025Failures to Surface Harmful Contents in Video Large Language Modelscited by 5
  9. 2026CurveStream: Boosting Streaming Video Understanding in MLLMs via Curvature-Aware Hierarchical Visual Memory Managementcited by 5
  10. 2026StreamOV: Streaming Omni-Video Understanding via Evidence-Guided Memory and Response Triggeringcited by 5
  11. 2026EGOSTREAM: A Diagnostic Benchmark for Streaming Episodic Memory in Egocentric Visioncited by 5
  12. 2026StreamTTT: Reconciling Real-Time Perception and Long-Term Memory in Streaming VLMscited by 5
  13. 2026Where to Look Matters: On-Policy Self-Distillation for Long-Video Understandingcited by 5
  14. 2026STRIDE: When to Speak Meets Sequence Denoising for Streaming Video Understandingcited by 4
  15. 2026StreamPro: From Reactive Perception to Proactive Decision-Making in Streaming Videocited by 4
  16. 2026An Efficient Streaming Video Understanding Framework with Agentic Controlcited by 4
  17. 2026Watch, Remember, Reason: Human-View Video Understanding with MLLMscited by 4
  18. 2026Harnessing Streaming Video in the Wildcited by 4
  19. 2026Keep It in Mind: User Centric Continual Spatial Intelligence Reasoning in Egocentric Video Streamscited by 4
  20. 2026Towards Spatial Supersensing in the Wildcited by 4
  21. 2025Delayed Bidirectional Alignment via Disentangled Audio Semantics for Audio-Visual Segmentationcited by 3
  22. 2026Question-guided Visual Compression with Memory Feedback for Long-Term Video Understandingcited by 3
  23. 2026Visual Agentic Memory: Enabling Online Long Video Understanding via Online Indexing, Hierarchical Memory, and Agentic Retrievalcited by 3
  24. 2026Omni-DuplexEval: Evaluating Real-time Duplex Omni-modal Interactioncited by 3
  25. 2026StreamScout: Learning When to Look Deeper for Streaming Video Understandingcited by 3
  26. 2025CFVBench: A Comprehensive Video Benchmark for Fine-grained Multimodal Retrieval-Augmented Generationcited by 2
  27. 2026RT-NeuS: Towards Real-Time Neuro-Symbolic Video Understanding via Adaptive Temporal Verificationcited by 2
  28. 2026Response-G1: Explicit Scene Graph Modeling for Proactive Streaming Video Understandingcited by 2
  29. 2026ShallowStream: Index Shallow then Answer Deep for Streaming Video Understandingcited by 2
  30. 2025CASA: Cross-Attention over Self-Attention for Efficient Vision-Language Fusioncited by 1
  31. 2026Going Down Memory Lane: Scaling Tokens for Video Stream Understanding with Dynamic KV-Cache Memorycited by 1
  32. 2026ReTool-Video: Recursive Tool-Using Video Agents with Meta-Augmented Tool Groundingcited by 1
  33. 2026CoRDS: Coreset-based Representative and Diverse Selection for Streaming Video Understandingcited by 1
  34. 2026Conditional Multi-Event Temporal Grounding in Long-Form Videocited by 1
  35. 2026ProtoKV: Streaming Video Understanding under Delayed Query with Summary-State Memorycited by 1
  36. 2026Aero Realtime: Fully Aligned Input-Output Streams for Low-Latency Streaming Multimodal Generationcited by 1
  37. 2026StreamEMS: Streaming Video Understanding with Self-Evolving Memory Scheme for Vision-Language Modelscited by 1
  38. 2026Beyond Retrieval: Progressive Latent Memory Evolution for Streaming Video Understandingcited by 1

50 most cited of 81 All 81 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026FlashVID: Efficient Video Large Language Models via Training-free Tree-based Spatiotemporal Token Merging
  2. 2024From Seconds to Hours: Reviewing MultiModal Large Language Models on Comprehensive Long Video Understanding
  3. 2024VideoChat-Flash: Hierarchical Compression for Long-Context Video Modeling
  4. 2026Event-VStream: Event-Driven Real-Time Understanding for Long Video Streams
  5. 2024Visual Context Window Extension: A New Perspective for Long Video Understanding
  6. 2024LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding
  7. 2026Scaling the Long Video Understanding of Multimodal Large Language Models via Visual Memory Mechanism
  8. 2026FlashGaze: Training-Free Multi-Scale Patch Pruning For Efficient Video Understanding