This paper introduces GPT4MTS, a prompt-based large language model for multimodal time series forecasting. The authors address the challenge of integrating textual information with numerical data to enhance forecasting performance. They propose a general pipeline for collecting textual information from various sources using modern large language models (LLMs) and create a GDELT-based multimodal time series dataset for news impact forecasting. The dataset includes both numerical time series data and textual summaries of events, providing a rich source of information for forecasting tasks. The GPT4MTS model leverages this multimodal input by using prompt tuning to guide the model in processing both types of data. Extensive experiments demonstrate the effectiveness of the proposed method, showing improved forecasting performance compared to unimodal models. The paper also discusses the importance of textual context and numerical precision in time series forecasting, highlighting the benefits of combining these modalities. The research contributes to enhancing communication accessibility and fostering further research in computational communication analysis.This paper introduces GPT4MTS, a prompt-based large language model for multimodal time series forecasting. The authors address the challenge of integrating textual information with numerical data to enhance forecasting performance. They propose a general pipeline for collecting textual information from various sources using modern large language models (LLMs) and create a GDELT-based multimodal time series dataset for news impact forecasting. The dataset includes both numerical time series data and textual summaries of events, providing a rich source of information for forecasting tasks. The GPT4MTS model leverages this multimodal input by using prompt tuning to guide the model in processing both types of data. Extensive experiments demonstrate the effectiveness of the proposed method, showing improved forecasting performance compared to unimodal models. The paper also discusses the importance of textual context and numerical precision in time series forecasting, highlighting the benefits of combining these modalities. The research contributes to enhancing communication accessibility and fostering further research in computational communication analysis.