Understanding time from visual representations is a fundamental cognitive skill, yet it remains a challenge for multimodal large language models (MLLMs). We investigate the capabilities of MLLMs in interpreting time and date through analogue clocks and yearly calendars. We curated ClockQA, comprising various types of clock styles paired with time-related questions, and CalendarQA, consisting of yearly calendar images with questions ranging from commonly known dates to computationally derived ones.

We aim to analyse how MLLMs can perform visual recognition, numerical reasoning, and temporal inference when presented with time-related visual data. Our evaluations show that despite recent advancements, reliably understanding time remains a significant challenge for MLLMs.