mlsanityData to 7 Oct 2026

InternVLA-A1: Unifying Understanding, Generation and Action for Robotic Manipulation

arXiv 2601.02456 ↗First posted 5 Jan 2026cs.ROcited by 48

On the map: Multimodal LLMs / vision-language models → Embodied and 3D vision-language models

also in 1 more
Abstract (arXiv)

Prevalent Vision-Language-Action (VLA) models are typically built upon Multimodal Large Language Models (MLLMs) and demonstrate exceptional proficiency in semantic understanding, but they inherently lack the capability to deduce physical world dynamics. Consequently, recent approaches have shifted toward World Models, typically formulated via video prediction; however, these methods often suffer from a lack of semantic grounding and exhibit brittleness in the presence of video prediction errors. To synergize semantic understanding with dynamic predictive capabilities, we present InternVLA-A1. This model employs a unified Mixture-of-Transformers architecture, coordinating three experts for scene understanding, visual foresight generation, and action execution. These components interact seamlessly through a unified masked self attention mechanism. Building upon InternVL3 and Qwen3-VL, we instantiate InternVLA-A1 at 2B and 3B parameter scales. We pre-train these models on heterogeneous data sources over real-world robot data, synthetic simulation data, and human videos, covering over 692M frames. This hybrid training strategy effectively harnesses the diversity of synthetic simulation data while minimizing the sim-to-real gap. We evaluated InternVLA-A1 on 12 real-world robotic tasks and a simulation benchmark. The results show that InternVLA-A1 consistently outperforms prior leading models: compared with pi0.5, it achieves +4.4% on static manipulation tasks and +2.6% on the RoboTwin 2.0 simulation benchmark, and delivers a +26.7% boost on dynamic manipulation tasks.

Key facts. Cited by 48 papers in the map (data through 7 Oct 2026). Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 72 of its 76 references that are in the map. First posted on arXiv 5 Jan 2026.

Builds on

72 of 76 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Igniting VLMs toward the Embodied Space15 Sep
  2. 2025CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models27 Mar
  3. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control28 Jul

Other references · 69

  1. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  3. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,337
  4. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  5. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasetscited by 2,526
  6. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  7. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,007
  8. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,653
  9. 2021Ego4D: Around the World in 3,000 Hours of Egocentric Videocited by 1,530
  10. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,167
  11. 2017The "something something" video database for learning and evaluating visual common sensecited by 1,092
  12. 2025Emerging Properties in Unified Multimodal Pretrainingcited by 855
38 more
  1. 2025Cosmos World Foundation Model Platform for Physical AIcited by 833
  2. 2024DROID: A Large-Scale In-The-Wild Robot Manipulation Datasetcited by 801
  3. 2025Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scalingcited by 787
  4. 2024Autoregressive Model Beats Diffusion: Llama for Scalable Image Generationcited by 760
  5. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 738
  6. 2023VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Modelscited by 733
  7. 2024PaliGemma: A versatile 3B VLM for transfercited by 679
  8. 2019RLBench: The Robot Learning Benchmark & Learning Environmentcited by 669
  9. 2024Open-Sora: Democratizing Efficient Video Production for Allcited by 637
  10. 2023BridgeData V2: A Dataset for Robot Learning at Scalecited by 634
  11. 2021CALVIN: A Benchmark for Language-Conditioned Policy Learning for Long-Horizon Robot Manipulation Taskscited by 564
  12. 2024LTX-Video: Realtime Video Latent Diffusioncited by 549
  13. 2023Learning Universal Policies via Text-Guided Video Generationcited by 506
  14. 2025RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulationcited by 477
  15. 2023Learning Interactive Real-World Simulatorscited by 440
  16. 2024RoboCasa: Large-Scale Simulation of Everyday Tasks for Generalist Robotscited by 391
  17. 2025AgiBot World Colosseo: A Large-scale Manipulation Platform for Scalable and Intelligent Embodied Systemscited by 390
  18. 2023Unleashing Large-Scale Video Generative Pre-training for Visual Robot Manipulationcited by 353
  19. 2025WorldVLA: Towards Autoregressive Action World Modelcited by 351
  20. 2023ManiSkill2: A Unified Benchmark for Generalizable Manipulation Skillscited by 313
  21. 2023MimicGen: A Data Generation System for Scalable Robot Learning using Human Demonstrationscited by 308
  22. 2024GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulationcited by 290
  23. 2025X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Modelcited by 285
  24. 2024Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representationscited by 279
  25. 2025Motus: A Unified Latent Action World Modelcited by 231
  26. 2020The EPIC-KITCHENS Dataset: Collection, Challenges and Baselinescited by 221
  27. 2022AnyGrasp: Robust and Efficient Grasp Perception in Spatial and Temporal Domainscited by 205
  28. 2025DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledgecited by 203
  29. 2025EgoDex: Learning Dexterous Manipulation from Large-Scale Egocentric Videocited by 191
  30. 2024RoboMIND: Benchmark on Multi-embodiment Intelligence Normative Data for Robot Manipulationcited by 166
  31. 2024Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulationcited by 159
  32. 2025Genie Envisioner: A Unified World Foundation Platform for Robotic Manipulationcited by 129
  33. 2025HAMSTER: Hierarchical Action Models For Open-World Robot Manipulationcited by 125
  34. 2025GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Datacited by 120
  35. 2025GR-3 Technical Reportcited by 110
  36. 2025OneTwoVLA: A Unified Vision-Language-Action Model with Adaptive Reasoningcited by 105
  37. 2025SANA-Sprint: One-Step Diffusion with Continuous-Time Consistency Distillationcited by 85
  38. 2025F1: A Vision-Language-Action Model Bridging Understanding and Generation to Actionscited by 85

50 most cited of 69 All 69 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 48.

48 papers.

Sort:

By area:

  1. 2026Cosmos 3: Omnimodal World Models for Physical AIcited by 137
  2. 2026VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Modelcited by 104
  3. 2026StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developingcited by 103
  4. 2026Qwen-RobotManip Technical Report: Alignment Unlocks Scale for Robotic Manipulation Foundation Modelscited by 47
  5. 2026World Action Models: The Next Frontier in Embodied AIcited by 46
  6. 2026World Model for Robot Learning: A Comprehensive Surveycited by 42
  7. 2026RoboDojo: A Unified Sim-and-Real Benchmark for Comprehensive Evaluation of Generalist Robot Manipulation Policiescited by 35
  8. 2026ReMem-VLA: Empowering Vision-Language-Action Model with Memory via Dual-Level Recurrent Queriescited by 28
  9. 2026StarVLA-α: Reducing Complexity in Vision-Language-Action Systemscited by 27
  10. 2026LangForce: Bayesian Decomposition of Vision Language Action Models via Latent Action Queriescited by 24
  11. 2026Omni-WorldBench: Towards a Comprehensive Interaction-Centric Evaluation for World Modelscited by 17
  12. 2026UniDriveVLA: Unifying Understanding, Perception, and Action Planning for Autonomous Drivingcited by 17
36 more
  1. 2026FutureVLA: Joint Visuomotor Prediction for Vision-Language-Action Modelcited by 14
  2. 2026HALO: A Unified Vision-Language-Action Model for Embodied Multimodal Chain-of-Thought Reasoningcited by 13
  3. 2026World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetrycited by 13
  4. 2026RoboInter: A Holistic Intermediate Representation Suite Towards Robotic Manipulationcited by 12
  5. 2026STARRY: Spatial-Temporal Action-Centric World Modeling for Robotic Manipulationcited by 10
  6. 2026Metis: A Generalizable and Efficient World-Action Model for Autonomous Driving and Urban Navigationcited by 10
  7. 2026Faster-WAM: Do World Action Models Need Deep Action Modules?cited by 9
  8. 2026Rethinking Visual-Language-Action Model Scaling: Alignment, Mixture, and Regularizationcited by 7
  9. 2026LaST-R1: Reinforcing Robotic Manipulation via Adaptive Physical Latent Reasoningcited by 7
  10. 2026Training Vision-Language-Action Models with Dense Embodied Chain-of-Thought Supervisioncited by 5
  11. 2026Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Modelscited by 5
  12. 2026Data Pyramid for Embodied Manipulation: A Surveycited by 4
  13. 2026LabVLA: Grounding Vision-Language-Action Models in Scientific Laboratoriescited by 3
  14. 2026From Imagined Futures to Executable Actions: Mixture of Latent Actions for Robot Manipulationcited by 2
  15. 2026Seeing Touch from Motion: A Unified Modality-Aware Visuo-Tactile Policy with Tactile Motion Correlationcited by 2
  16. 2026Bridge-WA: Predicting Where and How the World Changes for Robotic Actioncited by 2
  17. 2026FoMoVLA: Bridging Visual Foresight and Motion Guidance for Vision-Language-Action Modelscited by 2
  18. 2026N0-TWAM: Scaling Tactile-Native World-Action Model for Contact-Rich Manipulationcited by 2
  19. 2026PRM-as-a-Judge 1.5: A Toolkit for Robot Process Assessmentcited by 2
  20. 2026ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-trainingcited by 2
  21. 2026SIM1: Physics-Aligned Simulator as Zero-Shot Data Scaler in Deformable Worldscited by 1
  22. 2026RxBrain: Embodied Cognition Foundation Model with Joint Language-Visual Reasoning and Imaginationcited by 1
  23. 2026CometVLA: Co-Training on an Embodied Data Pyramid towards Physical Understandingcited by 1
  24. 2026InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Datacited by 1
  25. 2026EgoKit: Towards Unified Low-Cost Egocentric Data Collection with Heterogeneous Devicescited by 0
  26. 2026RhinoVLA Technical Reportcited by 0
  27. 2026Real-Time Execution with Autoregressive Policiescited by 0
  28. 2026N0-Foundation: Towards the Age of Tactile Intelligencecited by 0
  29. 2026DeCAL: Towards Physically-Grounded Dexterous Vision-Language-Action Models via Contact-Aware Latent Co-Imaginationcited by 0
  30. 2026WLA3: World Latent Action Modeling for Semantics, Dynamics, and Kinematicscited by 0
  31. 2026H-VLA: Hierarchical Vision-Language-Action Model with Key-Action Reasoning and Motion Planning in a Unified Action Spacecited by 0
  32. 2026RoboFL: Federated Expert Assembly for World Action Modelscited by 0
  33. 2026Rethinking Representations for World-Action Modelingcited by 0
  34. 2026EWAM: Emergent Depth-Wise Specialization in a Unified Embodied Model -- From Semantic Understanding through Visual Foresight to Actioncited by 0
  35. 2026TacDyn-WAM: Learning Implicit Tactile Dynamics in a Heterogeneous Visuo-Tactile World Action Modelcited by 0
  36. 2026Rewiring Semantics, Dynamics, and Control: A Simple yet Effective Action-Centric Tri-Stream Transformercited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026AWM-VLA: AlignedWorld Modeling for Efficient and Explainable Vision-Language-Action Policies
  2. 2025Vision Language Action Models in Robotic Manipulation: A Systematic Review
  3. 2025XR-1: Towards Versatile Vision-Language-Action Models via Learning Unified Vision-Motion Representations
  4. 2026InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization
  5. 2026SimVLA: A Simple VLA Baseline for Robotic Manipulation
  6. 2025Unified Vision-Language-Action Model
  7. 2026AC^2-VLA: Action-Context-Aware Adaptive Computation in Vision-Language-Action Models for Efficient Robotic Manipulation
  8. 2025HybridVLA: Collaborative Diffusion and Autoregression in a Unified Vision-Language-Action Model