How Fast Can I Run My VLA? Demystifying VLA Inference Performance with VLA-Perf

2026-02-20

AI Summary

Vision-Language-Action (VLA) models face stringent real-time inference demands in real-world robotic deployment, yet their performance is intricately coupled with both model architecture and system configuration, lacking systematic understanding. This work proposes VLA-Perf, the first performance model capable of analytically predicting end-to-end latency for arbitrary combinations of VLA models and inference systems. By integrating analytical modeling, multidimensional experimentation, and joint hardware-network simulation, this study systematically characterizes the VLA inference performance landscape for the first time. The analysis yields 15 key design principles spanning model scaling, architectural choices, long-context video processing, asynchronous inference, and edge-cloud collaboration strategies, offering actionable guidance for designing efficient VLA systems.

Technology Category

• Computer Vision: Large Vision Models
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Scalability of ML Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Vision-Language-Action (VLA) models have recently demonstrated impressive capabilities across various embodied AI tasks. While deploying VLA models on real-world robots imposes strict real-time inference constraints, the inference performance landscape of VLA remains poorly understood due to the large combinatorial space of model architectures and inference systems. In this paper, we ask a fundamental research question: How should we design future VLA models and systems to support real-time inference? To address this question, we first introduce VLA-Perf, an analytical performance model that can analyze inference performance for arbitrary combinations of VLA models and inference systems. Using VLA-Perf, we conduct the first systematic study of the VLA inference performance landscape. From a model-design perspective, we examine how inference performance is affected by model scaling, model architectural choices, long-context video inputs, asynchronous inference, and dual-system model pipelines. From the deployment perspective, we analyze where VLA inference should be executed -- on-device, on edge servers, or in the cloud -- and how hardware capability and network performance jointly determine end-to-end latency. By distilling 15 key takeaways from our comprehensive evaluation, we hope this work can provide practical guidance for the design of future VLA models and inference systems.

Problem

• Vision-Language-Action
• real-time inference
• inference performance
• embodied AI
• deployment constraints

Innovation

• VLA-Perf
• inference performance modeling
• real-time VLA inference
• systematic performance analysis
• embodied AI deployment