The integration of context-aware multimodal perception in Human-Robot Interaction (HRI) has emerged as a key factor in enhancing the naturalness, coherence, and adaptiveness of communication between humans and robots. This study explores how integrating a state of the art Visual Question Answering (VQA) model, Gemini, and Whisper for speech recognition within the humanoid robot Pepper impacts its communicative behavior and context-awareness attribution. Sixty participants interacted with Pepper in two conditions: one with access to contextual visual input and one without. Results revealed that context-aware interaction significantly enhanced user perceptions of Pepper’s ability to recognize human presence, behaviors and cognitions hence indicating an improved perception in human-aware capabilities. However, no statistically significant differences emerged in perceived context awareness across conditions evaluated by the use of ad-hoc questions, suggesting that anthropomorphic cues may contribute to users’ positive impressions regardless of actual system capabilities.