mlsanityData to 14 Oct 2026

InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency

arXiv 2508.18265 ↗First posted 25 Aug 2025cs.CVcited by 1,510

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

We introduce InternVL 3.5, a new family of open-source multimodal models that significantly advances versatility, reasoning capability, and inference efficiency along the InternVL series. A key innovation is the Cascade Reinforcement Learning (Cascade RL) framework, which enhances reasoning through a two-stage process: offline RL for stable convergence and online RL for refined alignment. This coarse-to-fine training strategy leads to substantial improvements on downstream reasoning tasks, e.g., MMMU and MathVista. To optimize efficiency, we propose a Visual Resolution Router (ViR) that dynamically adjusts the resolution of visual tokens without compromising performance. Coupled with ViR, our Decoupled Vision-Language Deployment (DvD) strategy separates the vision encoder and language model across different GPUs, effectively balancing computational load. These contributions collectively enable InternVL3.5 to achieve up to a +16.0% gain in overall reasoning performance and a 4.05× inference speedup compared to its predecessor, i.e., InternVL3. In addition, InternVL3.5 supports novel capabilities such as GUI interaction and embodied agency. Notably, our largest model, i.e., InternVL3.5-241B-A28B, attains state-of-the-art results among open-source MLLMs across general multimodal, reasoning, text, and agentic tasks -- narrowing the performance gap with leading commercial models like GPT-5. All models and code are publicly released.

Key facts. Cited by 1,510 papers in the map (data through 14 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 150 of its 183 references that are in the map. First posted on arXiv 25 Aug 2025.

Builds on

150 of 183 references.

Closest ancestors · 21

Same direction, newest first

  1. 2025Skywork-R1V3 Technical Report8 Jul
  2. 2025GTA1: GUI Test-time Scaling Agent8 Jul
  3. 2025BMMR: A Large-Scale Bilingual Multimodal Multi-Discipline Reasoning Dataset4 Jul
  4. 2025Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning23 Apr
  5. 2025VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning10 Apr
  6. 2025Kimi-VL Technical Report10 Apr
  7. 2025R1-Onevision: Advancing Generalized Multimodal Reasoning through Cross-Modal Formalization13 Mar
  8. 2025VisualPRM: An Effective Process Reward Model for Multimodal Reasoning13 Mar
  9. 2025MV-MATH: Evaluating Multimodal Math Reasoning in Multi-Visual Contexts28 Feb
  10. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  11. 2024Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction5 Dec
  12. 2024Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimization15 Nov
9 more
  1. 2024DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models29 Oct
  2. 2024CMM-Math: A Chinese Multimodal Math Dataset To Evaluate and Enhance the Mathematics Reasoning of Large Multimodal Models4 Sep
  3. 2024LogicVista: Multimodal LLM Logical Reasoning Benchmark in Visual Contexts6 Jul
  4. 2024We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?1 Jul
  5. 2024Benchmarking Multi-Image Understanding in Vision and Language Models: Perception, Knowledge, Reasoning, and Multi-Hop Reasoning18 Jun
  6. 2024Measuring Multimodal Mathematical Reasoning with MATH-Vision Dataset22 Feb
  7. 2023MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts3 Oct
  8. 2023MMBench: Is Your Multi-modal Model an All-around Player?12 Jul
  9. 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning19 Mar

Other references · 129

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2023GPT-4 Technical Reportcited by 18,396
  3. 2017Proximal Policy Optimization Algorithmscited by 15,215
  4. 2024The Llama 3 Herd of Modelscited by 14,715
  5. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  6. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  7. 2025Qwen3 Technical Reportcited by 8,820
  8. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  9. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  10. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  11. 2025Qwen2.5-VL Technical Reportcited by 5,587
  12. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
38 more
  1. 2024GPT-4o System Cardcited by 4,588
  2. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  3. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  4. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
  5. 2023Let's Verify Step by Stepcited by 3,789
  6. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,673
  7. 2023Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detectioncited by 3,332
  8. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  9. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,955
  10. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  11. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,810
  12. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,719
  13. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
  14. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
  15. 2023MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGIcited by 2,080
  16. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 2,057
  17. 2024Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameterscited by 1,986
  18. 2024MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmarkcited by 1,848
  19. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  20. 2025Gemma 3 Technical Reportcited by 1,824
  21. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,807
  22. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,729
  23. 2019Towards VQA Models That Can Readcited by 1,718
  24. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,652
  25. 2024Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysiscited by 1,413
  26. 2025s1: Simple test-time scalingcited by 1,341
  27. 2020DocVQA: A Dataset for VQA on Document Imagescited by 1,317
  28. 2023MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilitiescited by 1,239
  29. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  30. 2024OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environmentscited by 1,195
  31. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,136
  32. 2023MVBench: A Comprehensive Multi-modal Video Understanding Benchmarkcited by 1,109
  33. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  34. 2024How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suitescited by 1,050
  35. 2025Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?cited by 982
  36. 2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 928
  37. 2024Gemma: Open Models Based on Gemini Research and Technologycited by 916
  38. 2024Are We on the Right Way for Evaluating Large Vision-Language Models?cited by 914

50 most cited of 129 All 129 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 265; 2026: 1245.

1,510 papers.

Sort:

By area:

  1. 2026Cosmos 3: Omnimodal World Models for Physical AIcited by 165
  2. 2026Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Groundingcited by 139
  3. 2025MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsingcited by 116
  4. 2025Latent Visual Reasoningcited by 104
  5. 2026Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memorycited by 100
  6. 2025PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Modelcited by 89
  7. 2025Lumina-DiMOO: An Omni Diffusion Large Language Model for Multi-Modal Generation and Understandingcited by 88
  8. 2025Scaling Spatial Intelligence with Multimodal Foundation Modelscited by 82
  9. 2025Monet: Reasoning in Latent Visual Space Beyond Images and Languagecited by 82
  10. 2026MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interactioncited by 78
  11. 2026Being-H0.7: A Latent World-Action Model from Egocentric Videoscited by 75
  12. 2026DeepSeek-OCR 2: Visual Causal Flowcited by 74
38 more
  1. 2025Detect Anything via Next Point Predictioncited by 73
  2. 2026Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalizationcited by 66
  3. 2025ScaleCUA: Scaling Open-Source Computer Use Agents with Cross-Platform Datacited by 60
  4. 2026PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsingcited by 57
  5. 2025ThinkMorph: Emergent Properties in Multimodal Interleaved Chain-of-Thought Reasoningcited by 56
  6. 2025Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMscited by 55
  7. 2026Video-MME-v2: Towards the Next Stage in Benchmarks for Comprehensive Video Understandingcited by 55
  8. 2025OpenVE-3M: A Large-Scale High-Quality Dataset for Instruction-Guided Video Editingcited by 53
  9. 2025OneCAT: Decoder-Only Auto-Regressive Model for Unified Understanding and Generationcited by 50
  10. 2026MolmoAct2: Action Reasoning Models for Real-world Deploymentcited by 50
  11. 2026BabyVision: Visual Reasoning Beyond Languagecited by 49
  12. 2025Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agentscited by 46
  13. 2025G2VLM: Geometry Grounded Vision Language Model with Unified 3D Reconstruction and Spatial Reasoningcited by 44
  14. 2025How Far are VLMs from Visual Spatial Intelligence? A Benchmark-Driven Perspectivecited by 43
  15. 2025StreamingTOM: Streaming Token Compression for Efficient Video Understandingcited by 43
  16. 2025Step-GUI Technical Reportcited by 39
  17. 2026VLM4VLA: Revisiting Vision-Language-Models in Vision-Language-Action Modelscited by 39
  18. 2026LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligencecited by 39
  19. 2025VisMem: Latent Vision Memory Unlocks Potential of Vision-Language Modelscited by 37
  20. 2025Don't Blind Your VLA: Aligning Visual Representations for OOD Generalizationcited by 35
  21. 2025PAI-Bench: A Comprehensive Benchmark For Physical AIcited by 34
  22. 2025TUNA: Taming Unified Visual Representations for Native Unified Multimodal Modelscited by 34
  23. 2025SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learningcited by 33
  24. 2026STEP3-VL-10B Technical Reportcited by 32
  25. 2026SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecturecited by 32
  26. 2025InteractiveOmni: A Unified Omni-modal Model for Audio-Visual Multi-turn Dialoguecited by 29
  27. 2025RzenEmbed: Towards Comprehensive Multimodal Retrievalcited by 29
  28. 2025Active Video Perception: Iterative Evidence Seeking for Agentic Long Video Understandingcited by 29
  29. 2025Reasoning Within the Mind: Dynamic Multimodal Interleaving in Latent Spacecited by 28
  30. 2026FluxMem: Adaptive Hierarchical Memory for Streaming Video Understandingcited by 28
  31. 2026InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editingcited by 28
  32. 2026LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decodingcited by 28
  33. 2025AVoCaDO: An Audiovisual Video Captioner Driven by Temporal Orchestrationcited by 27
  34. 2025ReWatch-R1: Boosting Complex Video Reasoning in Large Vision-Language Models through Agentic Data Synthesiscited by 26
  35. 2025V-Thinker: Interactive Thinking with Imagescited by 26
  36. 2026VINO: A Unified Visual Generator with Interleaved OmniModal Contextcited by 26
  37. 2026GLM-OCR Technical Reportcited by 26
  38. 2025Skyra: AI-Generated Video Detection via Grounded Artifact Reasoningcited by 25

50 most cited of 1,510 All 1,510 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles
  2. 2025PeRL: Permutation-Enhanced Reinforcement Learning for Interleaved Vision-Language Reasoning
  3. 2025Learning Only with Images: Visual Reinforcement Learning with Reasoning, Rendering, and Visual Feedback
  4. 2026DeepVision-103K: A Visually Diverse, Broad-Coverage, and Verifiable Mathematical Dataset for Multimodal Reasoning
  5. 2026Insight-V++: Towards Advanced Long-Chain Visual Reasoning with Multimodal Large Language Models
  6. 2025VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use
  7. 2025Open Vision Reasoner: Transferring Linguistic Cognitive Behavior for Visual Reasoning
  8. 2025SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models