We present 3EED, the first large-scale benchmark for 3D visual grounding across vehicles, drones, and quadrupeds, with over 134K 3D objects and 25K human-verified expressions in diverse outdoor scenes.