Interactive storytelling games combining social robots and people have been extensively studied in HRI. Recent advancements in large language models have enabled innovative applications in interactive storytelling with social robots. This paper investigates the acceptability of a social robot, TABAN, as a storytelling companion for children by using a within-participants study. We implemented a collaborative visual storytelling game on TABAN based on images from the Bloom dataset, integrating LLaVA as a vision language model (VLM), GPT-3.5-Turbo as a large language model (LLM), Google speech recognition, and Ariana text-to-speech (TTS). Ten children participated in human and robot-led storytelling sessions, followed by a questionnaire-based evaluation on a 5-point Likert scale. A paired T-test revealed no significant differences between the two storytelling conditions (p = 0.936), indicating that TABAN performed similarly to a human storyteller. However, children found it easier to immerse themselves in stories told by humans (p = 0.003). Additional evaluations indicated high acceptance of TABAN, with positive perceptions of its social attributes. Furthermore, six out of ten children preferred storytelling with TABAN, which highlights its potential as an interactive storytelling companion.