Lost in Space? Vision-Language Models Struggle with Relative Camera Pose Estimation | Read Paper on Bytez