RotBench: Evaluating Multimodal Large Language Models on Identifying Image Rotation

We investigate to what extent Multimodal Large Language Models (MLLMs) can accurately identify the orientation of input images rotated 0deg, 90deg, 180deg, and 270deg. This task demands robust visual reasoning capabilities to detect rotational cues and contextualize spatial relationships within images, regardless of their orientation. To evaluate MLLMs on these abilities, we introduce RotBench, a 350-image manually-filtered benchmark comprising lifestyle, portrait, and landscape images. Despite the relatively simple nature of this task, we show that several state-of-the-art open and proprietary MLLMs, including GPT-5, o3, and Gemini-2.5-Pro, do not reliably identify rotation in input images. Providing models with auxiliary information -- including captions, depth maps, and more -- or using chain-of-thought prompting offers only small and inconsistent improvements. Our results indicate that most models are able to reliably identify right-side-up (0deg) images, while certain models are able to identify upside-down (180deg) images. None can reliably distinguish between 90deg and 270deg rotated images. Simultaneously showing the image rotated in different orientations leads to moderate performance gains for reasoning models, while a modified setup using voting improves the performance of weaker models. We further show that fine-tuning does not improve models' ability to distinguish 90deg and 270deg rotations, despite substantially improving the identification of 180deg images. Together, these results reveal a significant gap between MLLMs' spatial reasoning capabilities and human perception in identifying rotation.