mlsanityData to 14 Oct 2026

A Survey on Vision-Language-Action Models: An Action Tokenization Perspective

arXiv 2507.01925 ↗First posted 2 Jul 2025cs.ROcited by 96

On the map: Vision-language-action models → VLA architectures and generalization

Abstract (arXiv)

The remarkable advancements of vision and language foundation models in multimodal understanding, reasoning, and generation has sparked growing efforts to extend such intelligence to the physical world, fueling the flourishing of vision-language-action (VLA) models. Despite seemingly diverse approaches, we observe that current VLA models can be unified under a single framework: vision and language inputs are processed by a series of VLA modules, producing a chain of textitaction tokens that progressively encode more grounded and actionable information, ultimately generating executable actions. We further determine that the primary design choice distinguishing VLA models lies in how action tokens are formulated, which can be categorized into language description, code, affordance, trajectory, goal state, latent representation, raw action, and reasoning. However, there remains a lack of comprehensive understanding regarding action tokens, significantly impeding effective VLA development and obscuring future directions. Therefore, this survey aims to categorize and interpret existing VLA research through the lens of action tokenization, distill the strengths and limitations of each token type, and identify areas for improvement. Through this systematic review and analysis, we offer a synthesized outlook on the broader evolution of VLA models, highlight underexplored yet promising directions, and contribute guidance for future research, hoping to bring the field closer to general-purpose intelligence.

Key facts. Cited by 96 papers in the map (data through 14 Oct 2026). Placed in VLA architectures and generalization, one of 13 directions in Vision-language-action models. Builds on 97 of its 320 references that are in the map. First posted on arXiv 2 Jul 2025.

Builds on

97 of 320 references.

References · 97

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  4. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  5. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  6. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  7. 2020Scaling Laws for Neural Language Modelscited by 6,586
  8. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,585
  9. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  10. 2025Qwen2.5-VL Technical Reportcited by 5,587
  11. 2024GPT-4o System Cardcited by 4,588
  12. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
38 more
  1. 2013Auto-Encoding Variational Bayescited by 4,053
  2. 2021On the Opportunities and Risks of Foundation Modelscited by 4,033
  3. 2023Qwen Technical Reportcited by 3,344
  4. 2023Mistral 7Bcited by 3,164
  5. 2024SAM 2: Segment Anything in Images and Videoscited by 2,863
  6. 2020RAFT: Recurrent All-Pairs Field Transforms for Optical Flowcited by 2,415
  7. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,224
  8. 2024Gemma 2: Improving Open Language Models at a Practical Sizecited by 1,998
  9. 2019Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelismcited by 1,868
  10. 2025Gemma 3 Technical Reportcited by 1,824
  11. 2024Mixtral of Expertscited by 1,823
  12. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,761
  13. 2024OpenAI o1 System Cardcited by 1,244
  14. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,225
  15. 2024Grounded SAM: Assembling Open-World Models for Diverse Visual Taskscited by 929
  16. 2024Gemma: Open Models Based on Gemini Research and Technologycited by 916
  17. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 809
  18. 2024PaliGemma: A versatile 3B VLM for transfercited by 696
  19. 2024RDT-1B: a Diffusion Foundation Model for Bimanual Manipulationcited by 691
  20. 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 665
  21. 2023OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Modelscited by 565
  22. 2023VideoCrafter1: Open Diffusion Models for High-Quality Video Generationcited by 535
  23. 2020robosuite: A Modular Simulation Framework and Benchmark for Robot Learningcited by 511
  24. 2024Evaluating Real-World Robot Manipulation Policies in Simulationcited by 495
  25. 2018Universal Sentence Encodercited by 489
  26. 2025CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Modelscited by 451
  27. 2024CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulationcited by 422
  28. 2025AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systemscited by 405
  29. 2023An Embodied Generalist Agent in 3D Worldcited by 365
  30. 2024Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Modelscited by 329
  31. 2024Flow Matching Guide and Codecited by 301
  32. 2024GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulationcited by 296
  33. 2023AI Alignment: A Comprehensive Surveycited by 263
  34. 2023PaLI-X: On Scaling up a Multilingual Vision and Language Modelcited by 255
  35. 2021Bridge Data: Boosting Generalization of Robotic Skills with Cross-Domain Datasetscited by 250
  36. 2024EMMA: End-to-End Multimodal Model for Autonomous Drivingcited by 231
  37. 2025Hi Robot: Open-Ended Instruction Following with Hierarchical Vision-Language-Action Modelscited by 221
  38. 2025Real-Time Execution of Action Chunking Flow Policiescited by 216

50 most cited of 97 All 97 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 36; 2026: 60.

96 papers.

Sort:

By area:

  1. 2025MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulationcited by 192
  2. 2025A Survey of Reinforcement Learning for Large Reasoning Modelscited by 180
  3. 2025SimpleVLA-RL: Scaling VLA Training via Reinforcement Learningcited by 151
  4. 2025Vision-Language-Action Models for Robotics: A Review Towards Real-World Applicationscited by 84
  5. 2025WoW: Towards a World omniscient World model Through Embodied Interactioncited by 67
  6. 2025A Survey on Efficient Vision-Language-Action Modelscited by 34
  7. 2025VLSA: Vision-Language-Action Models with Plug-and-Play Safety Constraint Layercited by 34
  8. 2026DynamicVLA: A Vision-Language-Action Model for Dynamic Object Manipulationcited by 34
  9. 2025Vision-Language-Action Models for Autonomous Driving: Past, Present, and Futurecited by 33
  10. 2025VLA-Arena: An Open-Source Framework for Benchmarking Vision-Language-Action Modelscited by 31
  11. 2026QuantVLA: Scale-Calibrated Post-Training Quantization for Vision-Language-Action Modelscited by 30
  12. 2025Efficient Vision-Language-Action Models for Embodied Manipulation: A Systematic Surveycited by 26
38 more
  1. 2025RobustVLA: On Robustness of Vision-Language-Action Model against Multi-Modal Perturbationscited by 25
  2. 2025Hybrid Token Compression for Vision-Language Modelscited by 20
  3. 2026Show-Harness: Just a VLM Agent Can Play Robotscited by 20
  4. 2025DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Modelscited by 18
  5. 2025Seeing to Act, Prompting to Specify: A Bayesian Factorization of Vision Language Action Policycited by 18
  6. 2025SafeMind: Benchmarking and Mitigating Safety Risks in Embodied LLM Agentscited by 17
  7. 2025GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulationcited by 17
  8. 2026Joint-Aligned Latent Action: Towards Scalable VLA Pretraining in the Wildcited by 15
  9. 2025DualVLA: Building a Generalizable Embodied Agent via Partial Decoupling of Reasoning and Actioncited by 12
  10. 2025Learning Interactive World Model for Object-Centric Reinforcement Learningcited by 11
  11. 2026What Matters in Orchestrating Robot Policies: A Systematic Study of Hierarchical VLA Agentscited by 11
  12. 2025OpenHA: A Series of Open-Source Hierarchical Agentic Models in Minecraftcited by 10
  13. 2025Compose Your Policies! Improving Diffusion-based or Flow-based Robot Policies via Test-time Distribution-level Compositioncited by 9
  14. 2025Phantom Menace: Exploring and Enhancing the Robustness of VLA Models Against Physical Sensor Attackscited by 9
  15. 2025RoboOmni: Proactive Robot Manipulation in Omni-modal Contextcited by 8
  16. 2026Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVRcited by 8
  17. 2026ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Modelscited by 8
  18. 2025Survey of Vision-Language-Action Models for Embodied Manipulationcited by 7
  19. 2025HyperVLA: Efficient Inference in Vision-Language-Action Models via Hypernetworkscited by 7
  20. 2026RL-VLA3: A Flexible and Asynchronous Reinforcement Learning Framework for VLA Trainingcited by 7
  21. 2026H-WM: Robotic Task and Motion Planning Guided by Hierarchical World Modelcited by 7
  22. 2026Data Pyramid for Embodied Manipulation: A Surveycited by 6
  23. 2025Beyond Human Demonstrations: Diffusion-Based Reinforcement Learning to Generate Data for VLA Trainingcited by 5
  24. 2025SpatialActor: Exploring Disentangled Spatial Representations for Robust Robotic Manipulationcited by 5
  25. 2026APEX: A Decoupled Memory-based Explorer for Asynchronous Aerial Object Goal Navigationcited by 5
  26. 2026Efficient Long-Horizon Vision-Language-Action Models via Static-Dynamic Disentanglementcited by 5
  27. 2026ReconVLA: An Uncertainty-Guided and Failure-Aware Vision-Language-Action Framework for Robotic Controlcited by 5
  28. 2026Pre-VLA: Preemptive Runtime Verification for Reliable Vision-Language-Action and World-Model Rolloutscited by 5
  29. 2026APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policiescited by 5
  30. 2025PosA-VLA: Enhancing Action Generation via Pose-Conditioned Anchor Attentioncited by 4
  31. 2025An Anatomy of Vision-Language-Action Models: From Modules to Milestones and Challengescited by 4
  32. 2026E-VLA: Event-Augmented Vision-Language-Action Model for Dark and Blurred Scenescited by 4
  33. 2026Sword: Style-Robust World Models as Simulators via Dynamic Latent Bootstrapping for VLA Policy Post-Trainingcited by 4
  34. 2025Robotic Manipulation via Imitation Learning: Taxonomy, Evolution, Benchmark, and Challengescited by 3
  35. 2025Agentic Physical AI toward a Domain-Specific Foundation Model for Energy Systems: A Case Study on Nuclear Reactor Controlcited by 3
  36. 2026Vision-Language Models Unlock Task-Centric Latent Actionscited by 3
  37. 2026From Knowing to Doing Precisely: A General Self-Correction and Termination Framework for VLA modelscited by 3
  38. 2026Self-Supervised Bootstrapping of Action-Predictive Embodied Reasoningcited by 3

50 most cited of 96 All 96 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Vision-Language-Action (VLA) Models: Concepts, Progress, Applications and Challenges
  2. 202510 Open Challenges Steering the Future of Vision-Language-Action Models
  3. 2026Vision-Language-Action in Robotics: A Survey of Datasets, Benchmarks, and Data Engines
  4. 2026StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developing
  5. 2026ActionCodec: What Makes for Good Action Tokenizers
  6. 2025Vision Language Action Models in Robotic Manipulation: A Systematic Review
  7. 2025Unveiling the Potential of Vision-Language-Action Models with Open-Ended Multimodal Instructions
  8. 2026Not All Features Are Created Equal: A Mechanistic Study of Vision-Language-Action Models