VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Model

While this approach greatly enhances performance, it also incurs significant training costs. In this paper, we investigate how to effectively bridge vision-language (VL) representations to action (A). We introduce VLA-Adapter, a novel paradigm designed to reduce the reliance of VLA models on large-scale VLMs and extensive pre-training. To this end, we first systematically analyze the effectiveness of various VL conditions and present key findings on which conditions are essential for bridging perception and action spaces. Based on these insights, we propose a lightweight Policy module with Bridge Attention, which autonomously injects the optimal condition into the action space. In this way, our method achieves high performance using only a 0.5B-parameter backbone, without any robotic data pre-training. Extensive experiments on both simulated and real-world robotic benchmarks demonstrate that VLA-Adapter not only achieves state-of-the-art level performance, but also offers the fast inference speed reported to date. Furthermore, thanks to the proposed advanced bridging paradigm, VLA-Adapter enables the training of a powerful VLA model in just 8 hours on a single consumer-grade GPU, greatly lowering the barrier to deploying the VLA model. " " is that smaller values are better, and vice versa. Our paradigm can effectively obtain the SOT A-level VLA model using a tiny-scale backbone. In the past two years, with significant breakthroughs in multi-modal LLMs (Karamcheti et al., 2024; Steiner et al., 2024; Liu et al., 2023b; Li et al., 2025b), developing robot systems with general perception, understanding, and behavior capabilities has become a key research direction in artificial intelligence. In particular, the emergence of the Vision-Language-Action (VLA) model offers a new solution for enabling robot operations driven by instructions (Kim et al., 2024; Cui et al., 2025; Kim et al., 2025; Song et al., 2025b; Cen et al., 2025; Zhang et al., 2025b; Shi et al., 2025).