# Understanding Asynchronous Inference Methods for Vision-Language-Action Models

arXiv: https://arxiv.org/abs/2605.08168 · HTML: https://mlsanity.com/paper/2605.08168/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models.
- Builds on 12 of its 16 references that are in the map.
- First posted on arXiv 4 May 2026.

## Placed in

- Vision-language-action models / [Efficient VLA inference and adaptation](https://mlsanity.com/vla/efficient-vla-inference-and-adaptation/)

## Builds on (most cited first, 12 of 12)

- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [RT-1: Robotics Transformer for Real-World Control at Scale](https://mlsanity.com/paper/2212.06817/) (2022)
- [Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware](https://mlsanity.com/paper/2304.13705/) (2023)
- [MLP-Mixer: An all-MLP Architecture for Vision](https://mlsanity.com/paper/2105.01601/) (2021)
- [Real-Time Execution of Action Chunking Flow Policies](https://mlsanity.com/paper/2506.07339/) (2025)
- [Training-free Linear Image Inverses via Flows](https://mlsanity.com/paper/2310.04432/) (2023)
- [Training-Time Action Conditioning for Efficient Real-Time Chunking](https://mlsanity.com/paper/2512.05964/) (2025)
- [VLASH: Real-Time VLAs via Future-State-Aware Asynchronous Inference](https://mlsanity.com/paper/2512.01031/) (2025)
- [Leave No Observation Behind: Real-time Correction for VLA Action Chunks](https://mlsanity.com/paper/2509.23224/) (2025)
- [Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks](https://mlsanity.com/paper/2410.23208/) (2024)
- [Running VLAs at Real-time Speed](https://mlsanity.com/paper/2510.26742/) (2025)
- [How Fast Can I Run My VLA? Demystifying VLA Inference Performance with VLA-Perf](https://mlsanity.com/paper/2602.18397/) (2026)

## Built on by (most cited first, 4 of 4)

- [SmoothRL: Online Reinforcement Learning During Asynchronous Execution](https://mlsanity.com/paper/2608.29768/) (2026)
- [Catch Me If You Can: Real-Time Feedback Denoising for Responsive VLAs](https://mlsanity.com/paper/2609.21022/) (2026)
- [Staircase Policy: Streaming Inference for World-Action Models with Large Action Chunks](https://mlsanity.com/paper/2609.36471/) (2026)
- [Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation](https://mlsanity.com/paper/2609.39822/) (2026)

## Abstract (arXiv)

Vision-Language-Action (VLA) models offer a promising path to generalist robot control, but their inference latency causes observation staleness when generated actions are executed asynchronously. Several methods have been proposed concurrently to mitigate this problem: inference-time inpainting (IT-RTC), training-time delay simulation (TT-RTC), future-state-aware conditioning (VLASH), and lightweight residual correction (A2C2). Each takes a fundamentally different approach, but they have so far been evaluated independently with different codebases, base policies, and protocols. We present a systematic comparison of these four methods under controlled conditions. We develop two unified codebases that integrate all methods with harmonized library and dataset versions, and we benchmark them on the Kinetix suite with MLPMixer policies and on the LIBERO manipulation benchmark with SmolVLA, sweeping inference delays up to d=20 control steps. A2C2's per-step residual correction is the most effective method on Kinetix, holding above 90% solve rate up to d=8, and also leads on LIBERO from d=4 onwards. IT-RTC is competitive at low delays but degrades sharply under long chunks (H=30) and high delays. TT-RTC is the most robust training-based method: stable across d_max choices, generalizes beyond its training delay distribution, and adds zero inference overhead. VLASH exhibits a clear low-delay vs. high-delay trade-off governed by the fine-tuning delay range [0,d_max]. Code is available at https://github.com/TheAyos/async-vla-inference