StarVLA: Lego-like Codebase for VLA Model Development (April 2026)

StarVLA is an open-source, Lego-like codebase for Vision-Language-Action (VLA) model development, designed to address fragmentation across incompatible architectures, codebases, and evaluation protocols in embodied AI research. It provides a modular backbone–action-head architecture that supports both VLM backbones (e.g., Qwen-VL) and world-model backbones (e.g., Cosmos) alongside representative action-decoding paradigms, all under a shared abstraction that allows independent swapping of backbone and action head. The framework includes reusable training strategies such as cross-embodiment learning and multimodal co-training, and integrates major benchmarks—LIBERO, SimplerEnv, RoboTwin 2.0, RoboCasa-GR1, and BEHAVIOR-1K—through a unified evaluation interface supporting both simulation and real-robot deployment. StarVLA also ships simple, fully reproducible single-benchmark training recipes that match or surpass prior methods on multiple benchmarks. It is actively maintained and expanded by the StarVLA community.