mlsanityData to 14 Oct 2026

POINTS-Long: Adaptive Dual-Mode Visual Reasoning in MLLMs

arXiv 2604.11627 ↗First posted 13 Apr 2026cs.CVcited by 0

On the map: LLM reasoning → Multimodal reasoning

also in 1 more
Abstract (arXiv)

Multimodal Large Language Models (MLLMs) have recently demonstrated remarkable capabilities in cross-modal understanding and generation. However, the rapid growth of visual token sequences--especially in long-video and streaming scenarios--poses a major challenge to their scalability and real-world deployment. Thus, we introduce POINTS-Long, a native dual-mode MLLM featuring dynamic visual token scaling inspired by the human visual system. The model supports two complementary perception modes: focus mode and standby mode, enabling users to dynamically trade off efficiency and accuracy during inference. On fine-grained visual tasks, the focus mode retains the optimal performance, while on long-form general visual understanding, the standby mode retains 97.7-99.7% of the original accuracy using only 1/40-1/10th of the visual tokens. Moreover, POINTS-Long natively supports streaming visual understanding via a dynamically detachable KV-cache design, allowing efficient maintenance of ultra-long visual memory. Our work provides new insights into the design of future MLLMs and lays the foundation for adaptive and efficient long-form visual understanding.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 65 of its 125 references that are in the map. First posted on arXiv 13 Apr 2026.

Builds on

65 of 125 references.

Closest ancestors · 9

Same direction, newest first

  1. 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency25 Aug
  2. 2025Ovis2.5 Technical Report15 Aug
  3. 2025Thinking With Videos: Multimodal Tool-Augmented Reinforcement Learning for Long Video Reasoning6 Aug
  4. 2025MiMo-VL Technical Report4 Jun
  5. 2025Kimi-VL Technical Report10 Apr
  6. 2025Reason-RFT: Reinforcement Fine-Tuning for Visual Reasoning of Vision Language Models26 Mar
  7. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  8. 2024LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models22 Mar
  9. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts3 Oct

Other references · 56

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2025Qwen3 Technical Reportcited by 8,820
  3. 2025Qwen2.5-VL Technical Reportcited by 5,587
  4. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  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. 2019Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelismcited by 1,868
  8. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  9. 2023MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilitiescited by 1,239
  10. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  11. 2022Token Merging: Your ViT But Fastercited by 1,021
  12. 2025Emerging Properties in Unified Multimodal Pretrainingcited by 897
38 more
  1. 2023VideoChat: Chat-Centric Video Understandingcited by 857
  2. 2025Qwen2.5-Omni Technical Reportcited by 736
  3. 2023MiniGPT-v2: large language model as a unified interface for vision-language multi-task learningcited by 625
  4. 2024LLaVA-Video: Video Instruction Tuning With Synthetic Datacited by 583
  5. 2025VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understandingcited by 518
  6. 2024SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inferencecited by 391
  7. 2025Seed1.5-VL Technical Reportcited by 373
  8. 2024Mini-Gemini: Mining the Potential of Multi-modality Vision Language Modelscited by 365
  9. 2024TempCompass: Do Video LLMs Really Understand Videos?cited by 363
  10. 2024LMMs-Eval: Reality Check on the Evaluation of Large Multimodal Modelscited by 353
  11. 2024OpenVid-1M: A Large-Scale High-Quality Dataset for Text-to-video Generationcited by 341
  12. 2024PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reductioncited by 294
  13. 2024VideoChat-Flash: Hierarchical Compression for Long-Context Video Modelingcited by 220
  14. 2024Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understandingcited by 213
  15. 2025MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipecited by 186
  16. 2021SPViT: Enabling Faster Vision Transformers via Soft Token Pruningcited by 167
  17. 2025LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Tokencited by 166
  18. 2025WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMscited by 155
  19. 2025PaddleOCR 3.0 Technical Reportcited by 143
  20. 2024ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inferencecited by 134
  21. 2024CinePile: A Long Video Question Answering Dataset and Benchmarkcited by 114
  22. 2024TemporalBench: Benchmarking Fine-grained Temporal Understanding for Multimodal Video Modelscited by 97
  23. 2025Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memorycited by 97
  24. 2024LOOK-M: Look-Once Optimization in KV Cache for Efficient Multimodal Long-Context Inferencecited by 95
  25. 2025StreamingVLM: Real-Time Understanding for Infinite Video Streamscited by 95
  26. 2024CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understandingcited by 85
  27. 2025Flash-VStream: Efficient Real-Time Understanding for Long Video Streamscited by 81
  28. 2025Token Pruning in Multimodal Large Language Models: Are We Solving the Right Problem?cited by 76
  29. 2025Kwai Keye-VL 1.5 Technical Reportcited by 73
  30. 2024Multi-Stage Vision Token Dropping: Towards Efficient Multimodal Large Language Modelcited by 68
  31. 2024PruneVid: Visual Token Pruning for Efficient Video Large Language Modelscited by 67
  32. 2024Dynamic-LLaVA: Efficient Multimodal Large Language Models via Dynamic Vision-language Context Sparsificationcited by 59
  33. 2025FastVID: Dynamic Density Pruning for Fast Video Large Language Modelscited by 52
  34. 2025STORM: Token-Efficient Long Video Understanding for Multimodal LLMscited by 49
  35. 2025Video-XL-Pro: Reconstructive Token Compression for Extremely Long Video Understandingcited by 49
  36. 20251.4 Million Open-Source Distilled Reasoning Dataset to Empower Large Language Model Trainingcited by 49
  37. 2024AIM: Adaptive Inference of Multi-Modal LLMs via Token Merging and Pruningcited by 43
  38. 2025VisualWebInstruct: Scaling up Multimodal Instruction Data through Web Searchcited by 35

50 most cited of 56 All 56 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Accelerating Multimodal Large Language Models via Dynamic Visual-Token Exit and the Empirical Findings
  2. 2025Multi-Step Visual Reasoning with Visual Tokens Scaling and Verification
  3. 2026LookWise: Knowing When and Where to Look for Fine-Grained Visual Reasoning in Multimodal Large Language Models
  4. 2026Vision Inference Former: Sustaining Visual Consistency in Multimodal Large Language Models
  5. 2023LION : Empowering Multimodal Large Language Model with Dual-Level Visual Knowledge
  6. 2025Look-Back: Implicit Visual Re-focusing in MLLM Reasoning
  7. 2026Insight-V++: Towards Advanced Long-Chain Visual Reasoning with Multimodal Large Language Models
  8. 2026Towards Efficient Large Vision-Language Models: A Comprehensive Survey on Inference Strategies