The first decision is not price but control authority. The Unitree G1 Basic can support demonstrations, interaction experiments, and bounded internal programs, yet organizations often underestimate how quickly a closed or restricted development surface becomes the dominant constraint once the project moves beyond choreographed motion. A team that expects to modify locomotion behavior, integrate external perception, build ROS2 pipelines, collect synchronized telemetry, or test simulation-to-real transfer needs secondary-development access from the outset–purchasing a lower tier and later discovering that the control stack cannot support the research roadmap creates a more expensive failure than buying excess capability.