mlsanityData to 14 Oct 2026

Kwai Keye-VL 1.5 Technical Report

arXiv 2509.01563 ↗First posted 1 Sep 2025cs.CVcited by 73

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

Abstract (arXiv)

In recent years, the development of Large Language Models (LLMs) has significantly advanced, extending their capabilities to multimodal tasks through Multimodal Large Language Models (MLLMs). However, video understanding remains a challenging area due to the dynamic and information-dense nature of videos. Existing models struggle with the trade-off between spatial resolution and temporal coverage when processing video content. We present Keye-VL-1.5, which addresses fundamental challenges in video comprehension through three key innovations. First, we introduce a novel Slow-Fast video encoding strategy that dynamically allocates computational resources based on inter-frame similarity, processing key frames with significant visual changes at higher resolution (Slow pathway) while handling relatively static frames with increased temporal coverage at lower resolution (Fast pathway). Second, we implement a progressive four-stage pre-training methodology that systematically extends the model's context length from 8K to 128K tokens, enabling processing of longer videos and more complex visual content. Third, we develop a comprehensive post-training pipeline focusing on reasoning enhancement and human preference alignment, incorporating a 5-step chain-of-thought data construction process, iterative GSPO-based reinforcement learning with progressive prompt hinting for difficult cases, and alignment training. Through extensive evaluation on public benchmarks and rigorous internal human assessment, Keye-VL-1.5 demonstrates significant improvements over existing models, particularly excelling in video understanding tasks while maintaining competitive performance on general multimodal benchmarks.

Key facts. Cited by 73 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 76 of its 90 references that are in the map. First posted on arXiv 1 Sep 2025.

Builds on

76 of 90 references.

Closest ancestors · 6

Same direction, newest first

  1. 2025TEMPURA: Temporal Event Masked Prediction and Understanding for Reasoning in Action2 May
  2. 2025Long-VITA: Scaling Large Multi-modal Models to 1 Million Tokens with Leading Short-Context Accuracy7 Feb
  3. 2025Tarsier2: Advancing Large Vision-Language Models from Detailed Video Description to Comprehensive Video Understanding14 Jan
  4. 2024LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding22 Jul
  5. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis31 May
  6. 2023Video-LLaVA: Learning United Visual Representation by Alignment Before Projection16 Nov

Other references · 70

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  4. 2024The Llama 3 Herd of Modelscited by 14,715
  5. 2025Qwen3 Technical Reportcited by 8,820
  6. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  7. 2025Qwen2.5-VL Technical Reportcited by 5,587
  8. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  9. 2024GPT-4o System Cardcited by 4,588
  10. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  11. 2016Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotationscited by 2,656
  12. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
38 more
  1. 2023MMBench: Is Your Multi-modal Model an All-around Player?cited by 2,141
  2. 2023MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGIcited by 2,080
  3. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,917
  4. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  5. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contextscited by 1,614
  6. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,599
  7. 2021Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Conceptscited by 1,216
  8. 2022Reproducible scaling laws for contrastive language-image learningcited by 1,173
  9. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,136
  10. 2024Are We on the Right Way for Evaluating Large Vision-Language Models?cited by 914
  11. 2016A Diagram Is Worth A Dozen Imagescited by 804
  12. 2024Emu3: Next-Token Prediction is All You Needcited by 781
  13. 2025Qwen2.5-Omni Technical Reportcited by 736
  14. 2024Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMscited by 735
  15. 2023DataComp: In search of the next generation of multimodal datasetscited by 644
  16. 2024MathVerse: Does Your Multi-modal LLM Truly See the Diagrams in Visual Math Problems?cited by 586
  17. 2024LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Modelscited by 584
  18. 2023HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Modelscited by 583
  19. 2024BLINK: Multimodal Large Language Models Can See but Not Perceivecited by 550
  20. 2024Measuring Multimodal Mathematical Reasoning with MATH-Vision Datasetcited by 526
  21. 2023OCRBench: On the Hidden Mystery of OCR in Large Multimodal Modelscited by 476
  22. 2025Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Modelcited by 461
  23. 2025Kimi-VL Technical Reportcited by 388
  24. 2025Seed1.5-VL Technical Reportcited by 373
  25. 2024TempCompass: Do Video LLMs Really Understand Videos?cited by 363
  26. 2024We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?cited by 286
  27. 2024LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contextscited by 245
  28. 2024Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimizationcited by 239
  29. 2025Video-MMMU: Evaluating Knowledge Acquisition from Multi-Discipline Professional Videoscited by 236
  30. 2023Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolutioncited by 229
  31. 2025VITA-1.5: Towards GPT-4o Level Real-Time Vision and Speech Interactioncited by 197
  32. 2025Thyme: Think Beyond Imagescited by 167
  33. 2025OpenThinkIMG: Learning to Think with Images via Visual Tool Reinforcement Learningcited by 163
  34. 2025Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learningcited by 161
  35. 2023RAIN: Your Language Models Can Align Themselves without Finetuningcited by 157
  36. 2024DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Modelscited by 146
  37. 2024Video-RAG: Visually-aligned Retrieval-Augmented Long Video Comprehensioncited by 138
  38. 2025MMVU: Measuring Expert-Level Multi-Discipline Video Understandingcited by 137

50 most cited of 70 All 70 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 14; 2026: 59.

73 papers.

Sort:

By area:

  1. 2026Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Groundingcited by 139
  2. 2026Video-MME-v2: Towards the Next Stage in Benchmarks for Comprehensive Video Understandingcited by 55
  3. 2025Spatia: Video Generation with Updatable Spatial Memorycited by 40
  4. 2026LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligencecited by 39
  5. 2025MathCanvas: Intrinsic Visual Chain-of-Thought for Multimodal Mathematical Reasoningcited by 35
  6. 2026InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editingcited by 28
  7. 2026LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decodingcited by 28
  8. 2025Streaming Video Instruction Tuningcited by 26
  9. 2025A Survey on Agentic Multimodal Large Language Modelscited by 23
  10. 2025CodePlot-CoT: Mathematical Visual Reasoning by Thinking with Code-Driven Imagescited by 23
  11. 2026AURA: Always-On Understanding and Real-Time Assistance via Video Streamscited by 16
  12. 2026MolmoPoint: Better Pointing for VLMs with Grounding Tokenscited by 15
38 more
  1. 2025UniRec-0.1B: Unified Text and Formula Recognition with 0.1B Parameterscited by 13
  2. 2026VideoVeritas: AI-Generated Video Detection via Perception Pretext Reinforcement Learningcited by 11
  3. 20254DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Modelscited by 10
  4. 2025Video-LMM Post-Training: A Deep Dive into Video Reasoning with Large Multimodal Modelscited by 8
  5. 2026PatchCue: Enhancing Vision-Language Model Reasoning with Patch-Based Visual Cuescited by 8
  6. 2026ResAdapt: Adaptive Resolution for Efficient Multimodal Reasoningcited by 8
  7. 2026Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoderscited by 7
  8. 2026IPIBench: Evaluating Interactive Proactive Intelligence of MLLMs under Continuous Streamscited by 7
  9. 2026AdaCodec: A Predictive Visual Code for Video MLLMscited by 7
  10. 2025IF-Bench: Benchmarking and Enhancing MLLMs for Infrared Images with Generative Visual Promptingcited by 6
  11. 2025Masking Teacher and Reinforcing Student for Distilling Vision-Language Modelscited by 6
  12. 2025TimeScope: Towards Task-Oriented Temporal Grounding In Long Videoscited by 5
  13. 2025A Benchmark for Omni-Modal Reasoning in Long Videoscited by 5
  14. 2026ContextRL: Enhancing MLLM's Knowledge Discovery Efficiency with Context-Augmented RLcited by 5
  15. 2026Small Vision-Language Models are Smart Compressors for Long Video Understandingcited by 5
  16. 2026Visual Preference Optimization with Rubric Rewardscited by 5
  17. 2026Moment-Video: Diagnosing Temporal Fidelity of Video MLLMs on Momentary Visual Eventscited by 5
  18. 2026Kwai Keye-VL-2.0 Technical Reportcited by 5
  19. 2026InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoningcited by 5
  20. 2025CARE What Fails: Contrastive Anchored-REflection for Verifiable Multimodal Reasoningcited by 4
  21. 2026VidVec: Unlocking Video MLLM Embeddings for Video-Text Retrievalcited by 4
  22. 2026Scaling Video Understanding via Compact Latent Multi-Agent Collaborationcited by 4
  23. 2026Watch, Remember, Reason: Human-View Video Understanding with MLLMscited by 4
  24. 2026Beyond Accuracy: Evaluating Grounded Visual Evidence in Thinking with Imagescited by 3
  25. 2026Joint Reward Modeling: Internalizing Chain-of-Thought for Efficient Visual Reward Modelscited by 3
  26. 2026DiningBench: A Hierarchical Multi-view Benchmark for Perception and Reasoning in the Dietary Domaincited by 3
  27. 2026Building a Precise Video Language with Human-AI Oversightcited by 3
  28. 2026SciVQR: A Multidisciplinary Multimodal Benchmark for Advanced Scientific Reasoning Evaluationcited by 3
  29. 2025The Potential and Limitations of Vision-Language Models for Human Motion Understanding: A Case Study in Data-Driven Stroke Rehabilitationcited by 2
  30. 2026QARM V2: Quantitative Alignment Multi-Modal Recommendation for Reasoning User Sequence Modelingcited by 2
  31. 2026Spatial Chain-of-Thought: Bridging Understanding and Generation Models for Spatial Reasoning Generationcited by 2
  32. 2026VirtueBench: Evaluating Trustworthiness under Uncertainty in Long Video Understandingcited by 2
  33. 2026DataComp-VLM: Improved Open Datasets for Vision-Language Modelscited by 2
  34. 2026MVP: Enhancing Video Large Language Models via Self-supervised Masked Video Predictioncited by 1
  35. 2026Kelix Technical Reportcited by 1
  36. 2026Chatting with Images for Introspective Visual Thinkingcited by 1
  37. 2026EXPLORE-Bench: Egocentric Scene Prediction with Long-Horizon Reasoningcited by 1
  38. 2026ESOM: Efficiently Understanding Streaming Video Anomalies with Open-world Dynamic Definitionscited by 1

50 most cited of 73 All 73 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Adaptive Keyframe Sampling for Long Video Understanding
  2. 2026KTV: Keyframes and Key Tokens Selection for Efficient Training-Free Video LLMs
  3. 2026Evaluation of MLLM-Agnostic Plug-and-Play Keyframe Selection Methods for Long Video Understanding
  4. 2025Threading Keyframe with Narratives: MLLMs as Strong Long Video Comprehenders
  5. 2024From Seconds to Hours: Reviewing MultiModal Large Language Models on Comprehensive Long Video Understanding
  6. 2025BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding
  7. 2024LongVLM: Efficient Long Video Understanding via Large Language Models
  8. 2025Video Panels for Long Video Understanding