ReactVLA: Fast and Lightweight Reactive Robot Manipulation via Improved Mean Flow Action Generation

Diffusion-based Vision-Language-Action (VLA) policies have demonstrated strong capability in modeling expressive and multimodal action distributions. However, their reliance on iterative sampling introduces substantial inference latency, which limits their applicability to reactive closed-loop robot manipulation. To address this limitation, we propose textttReactVLA, a lightweight and low-latency VLA framework for real-time robotic manipulation. textttReactVLA combines two complementary designs: (1) an improved Mean Flow (iMF) action generator that reduces expensive multi-step diffusion sampling to one-to-few-step action generation, and (2) Attention Residuals (AttnRes), a dynamic depth-wise feature routing mechanism that replaces uniform residual accumulation to better preserve task-relevant multimodal representations. We evaluate textttReactVLA on large-scale simulation benchmarks, including LIBERO and RoboIMI, as well as real-world robotic manipulation tasks. Experimental results show that textttReactVLA consistently outperforms similarly sized VLA baselines, including SmolVLA and π0. On challenging precision manipulation tasks, textttReactVLA achieves up to a 1.65× improvement in task performance while providing more than a 4× increase in inference speed compared with leading VLA models. Finally, it reduces real-world policy latency to below 38.6 ms, enabling fast reactive control on physical robot platforms. Please check out our project website at: https://game-loader.github.io/ReactVLA/.