| InternW0-Δ: A World Action Model Bridging Predictive Dynamics and Actions with 20K+ Hours of Open Data | | 92.8 | InternW0-Δ | 2026-09-25 |
| Qwen-RobotManip Technical Report: Alignment Unlocks Scale for Robotic Manipulation Foundation Models | ✓ Link | 91.4 | Qwen-RobotManip-Context | 2026-06-17 |
| REKI: Registering Knowledge in VLAs | ✓ Link | 90.4 | REKI | 2026-10-15 |
| Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment | ✓ Link | 90.3 | Anchor-Align VLA | 2026-07-15 |
| UNITAS: A 3D-Native World Action Model for Embodied Manipulation | ✓ Link | 89.1 | UNITAS | 2026-10-08 |
| WAVE-0: Bridging VLA Policies and WAMs with Lightweight Latent Future Prediction | ✓ Link | 88.0 | WAVE-0 | 2026-10-13 |
| Beyond Future Prediction: Denoising as Generative Adaptation for Robot Control | ✓ Link | 87.8 | NowWAM (Z-Image-6B) | 2026-09-23 |
| E-TTS: A New Embodied Test-Time Scaling Framework for Robotic Manipulation | | 87.15 | MolmoAct + E-TTS (zero-shot) | 2026-06-25 |
| Breaking the Vision-Action Shortcut: Latent Interface Training for Generalizable Robotics Foundation Models | ✓ Link | 86.89 | ImageWAM + LIT | 2026-09-11 |
| GreenWAM: A World Foundation Model for Robotics | ✓ Link | 86.8 | GreenWAM | 2026-10-13 |
| ACoT-VLA: Action Chain-of-Thought for Vision-Language-Action Models | ✓ Link | 86.6 | ACoT-VLA | 2026-01-16 |
| World Action Modeling with Progressive Visual Planning | | 85.8 | ProWAM (zero-shot) | 2026-10-01 |
| Geometric Action Model for Robot Policy Learning | ✓ Link | 85.5 | GAM | 2026-06-15 |
| VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse | ✓ Link | 85.42 | VLA-Rolling (pi0.5) | 2026-10-18 |
| InternVLA-A1.5: Unifying Understanding, Latent Foresight, and Action for Compositional Generalization | ✓ Link | 84.8 | InternVLA-A1.5-Libero | 2026-07-06 |
| World Pilot: Steering Vision-Language-Action Models with World-Action Priors | ✓ Link | 84.7 | World Pilot | 2026-06-10 |
| Learning the 4D World through Point Trajectory Forecasting | ✓ Link | 84.6 | VPM (V+T+A) (extra data) | 2026-10-20 |
| Vision-Language-Delta-Action: Learning Compact Deltas as Actionable Transition Interfaces for Robotic Manipulation | ✓ Link | 84.5 | LoongForge-VLDA | 2026-10-19 |
| Native Action-Prior Learning from Videos for World Action Models | | 83.5 | NAVA-WAM | 2026-10-02 |
| TBD-VLA: Temporal Block Diffusion Vision Language Action Model | | 83.49 | TBD-VLA (w/ pre-training) | 2026-06-05 |
| ABot-M0.5: Unified Mobility-and-Manipulation World Action Model | | 83.4 | ABot-M0.5 | 2026-07-01 |
| CorridorVLA: Explicit Spatial Constraints for Generative Action Heads via Sparse Anchors | | 83.21 | GR00T-Corr | 2026-04-23 |
| ImageWAM: Do World Action Models Really Need Video Generation, or Just Image Editing? | ✓ Link | 83.1 | ImageWAM (FLUX.2 4B) | 2026-06-17 |
| MM-ABC: Towards Generalist Mobile Manipulation via Seeing, Coordinating and Imagining | ✓ Link | 82.8 | MM-ABC | 2026-09-28 |
| Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models | ✓ Link | 82.6 | VLAct (PI) | 2026-08-27 |
| Acting Through Registers | ✓ Link | 81.64 | RegisterWAM | 2026-10-13 |
| PINT: A Unified Model with Shared Predictive Intent for Robot Manipulation | ✓ Link | 81.0 | PINT-VA | 2026-10-12 |
| Flex-$π$: A Multi-Stream World-Action Model with Compute Flexibility | ✓ Link | 80.9 | Flex-π (full joint) | 2026-08-11 |
| ABot-M0: VLA Foundation Model for Robotic Manipulation with Action Manifold Learning | ✓ Link | 80.5 | ABot-M0 | 2026-02-11 |
| GE-Act 2.0: Pretraining and Scaling a World-Action Model for Robotic Manipulation | | 80.4 | GE-Act2.0 | 2026-09-04 |
| VLANeXt: Recipes for Building Strong VLA Models | ✓ Link | 80.1 | VLANeXt | 2026-02-20 |
| Less is More: Reducing Complexity in Vision-Language-Action Systems | ✓ Link | 79.7 | StarVLA-α (Generalist, zero-shot) | 2026-09-08 |
| VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model | ✓ Link | 79.5 | VLA-JEPA | 2026-02-10 |
| What Makes World Action Models Generalize? An Empirical Study of Test-Time Future Modeling | ✓ Link | 79.5 | Simple-WAM | 2026-09-29 |
| LaMP: Learning Vision-Language-Action Policies with 3D Scene Flow as Latent Motion Prior | ✓ Link | 79.3 | LaMP | 2026-09-08 |
| DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models | ✓ Link | 79.0 | DeepThinkVLA (pi0-FAST) | 2025-10-31 |
| SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models | ✓ Link | 78.4 | SimpleMemVLA | 2026-09-02 |
| Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models? | ✓ Link | 77.6 | π0.5 + DTR (Drop-9) | 2026-06-26 |
| Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models | ✓ Link | 76.0 | Embodied-R1.5-VLA | 2026-06-09 |
| Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVR | ✓ Link | 75.93 | Pion (VLANeXt) | 2026-05-19 |
| Latent Action as Intention Enables Efficient Future Imagination for World Action Models | ✓ Link | 74.4 | LAWA (zero-shot) | 2026-08-25 |
| SG-WAM: Self-Guided World Modeling in Geometry-Aware Policy Space | ✓ Link | 73.0 | SG-WAM (0.9B) | 2026-08-02 |
| Dynin-Robotics: Omnimodal Unified Diffusion Vision-Language-Action Model | ✓ Link | 73.0 | Dynin-Robotics | 2026-09-11 |
| RealtimeWAM: One-Step Asynchronous World Action Models | ✓ Link | 73.0 | RealtimeWAM† | 2026-10-05 |
| TacitVLA: Reasoning Vision-Language-Action Models Without Reasoning Latency | | 72.9 | TacitVLA | 2026-10-13 |
| ST-WAM: Semantic-Temporal World Action Model for Robust Manipulation under Visual Distribution Shifts | ✓ Link | 72.8 | ST-WAM | 2026-07-31 |
| BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries | ✓ Link | 72.2 | BridgeWAM | 2026-10-10 |
| APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies | ✓ Link | 71.6 | APT | 2026-06-10 |
| OpenWAM: An Open, Modular Exploration Towards Systematic World-Action Model Pretraining | ✓ Link | 69.2 | OpenWAM-alpha (extra data) | 2026-09-07 |
| ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models | ✓ Link | 68.8 | ActionPiece + Qwen3-VL-4B (zero-shot) | 2026-09-16 |
| PAIR: Bridging Perception and Action in Vision-Language-Action Models | | 64.2 | VLA-Adapter + PAIR | 2026-10-06 |
| Can World and Action Modeling Benefit Each Other under Unified Causal Training? | ✓ Link | 59.27 | OneWM | 2026-10-15 |
| CorridorVLA: Explicit Spatial Constraints for Generative Action Heads via Sparse Anchors | | 57.74 | SmolVLA-Corr | 2026-04-23 |