New vision-language-action model demonstrates strong cross-morphology performance and enables scalable deployment across real-world robotic platforms