FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference

Summary

Abstract Summary

Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet their real-world deployment remains bottlenecked by high inference latency and unstable asynchronous execution. This challenge is particularly pronounced in flow-matching-based VLA models, where action decoding requires multiple iterative steps conditioned on the VLM context. While efficient inference methods improve control frequency and asynchronous methods reduce execution idle time, existing approaches often fail to jointly achieve low-latency inference and accurate, temporally consistent asynchronous execution. We introduce \textbf{FlashVLA}, a streaming action decoding framework that addresses both challenges in a unified formulation. FlashVLA maintains a streaming action buffer with multiple chunks at different noise levels and decodes them using chunk-wise causal attention. This design allows FlashVLA to produce one executable action chunk per inference step. Moreover, its chunk-wise autoregressive formulation implicitly preserves action continuity, enabling smooth asynchronous execution without extra future-state conditioning. Across extensive simulated and real-world experiments, FlashVLA substantially improves inference speed while maintaining strong task performance. It can achieve $\geq$30\,Hz control frequency on a single GPU with smooth asynchronous inference in real-world deployment.

Key Points

Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet thei...
This challenge is particularly pronounced in flow-matching-based VLA models, where action decodin...
While efficient inference methods improve control frequency and asynchronous methods reduce execu...
We introduce \textbf{FlashVLA}, a streaming action decoding framework that addresses both challen...
FlashVLA maintains a streaming action buffer with multiple chunks at different noise levels and d...

FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference

|Authors: Zekai Li, Jiaming Tang, Zhijian Liu

|Venue: arXiv preprint | Year: 2026

|arXiv: 2608.27384v1

Abstract

Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet their real-world deployment remains bottlenecked by high inference latency and unstable asynchronous execution. This challenge is particularly pronounced in flow-matching-based VLA models, where action decoding requires multiple iterative steps conditioned on the VLM context. While efficient inference methods improve control frequency and asynchronous methods reduce execution idle time, existing approaches often fail to jointly achieve low-latency inference and accurate, temporally consistent asynchronous execution. We introduce \textbf{FlashVLA}, a streaming action decoding framework that addresses both challenges in a unified formulation. FlashVLA maintains a streaming action buffer with multiple chunks at different noise levels and decodes them using chunk-wise causal attention. This design allows FlashVLA to produce one executable action chunk per inference step. Moreover, its chunk-wise autoregressive formulation implicitly preserves action continuity, enabling smooth asynchronous execution without extra future-state conditioning. Across extensive simulated and real-world experiments, FlashVLA substantially improves inference speed while maintaining strong task performance. It can achieve $\geq$30,Hz control frequency on a single GPU with smooth asynchronous inference in real-world deployment.

Key Contributions

Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet thei…
This challenge is particularly pronounced in flow-matching-based VLA models, where action decodin…
While efficient inference methods improve control frequency and asynchronous methods reduce execu…
We introduce \textbf{FlashVLA}, a streaming action decoding framework that addresses both challen…
FlashVLA maintains a streaming action buffer with multiple chunks at different noise levels and d…

Topics

manipulation
vision
vla
reinforcement-learning
control

Code & Data

GitHub repository: https://github.com/z-lab/flashvla.git

BibTeX

@article{Li2026_260827384v1,
  title     = {FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference},
  author    = {Zekai Li and Jiaming Tang and Zhijian Liu},
  year      = {2026},
  eprint    = {2608.27384v1},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  url       = {https://arxiv.org/abs/2608.27384v1}
}