The paper introduces the Mamba architecture, a state space model-based approach, for remote sensing change detection (CD) tasks, including binary change detection (BCD), semantic change detection (SCD), and building damage assessment (BDA). The Mamba architecture is designed to overcome the limitations of traditional Convolutional Neural Networks (CNNs) and Transformers, which struggle with capturing long-range dependencies and computational efficiency, respectively. The authors propose three specific frameworks, MambaBCD, MambaSCD, and MambaBDA, tailored for each CD task. These frameworks utilize the Visual Mamba architecture as the encoder to extract global spatial contextual information from multi-temporal images. For the change decoder, three spatio-temporal relationship modeling mechanisms are introduced to effectively capture spatio-temporal interactions in multi-temporal features. The proposed frameworks achieve competitive performance on five benchmark datasets without complex training strategies, demonstrating the potential of the Mamba architecture in CD tasks. The source code is available for public use.The paper introduces the Mamba architecture, a state space model-based approach, for remote sensing change detection (CD) tasks, including binary change detection (BCD), semantic change detection (SCD), and building damage assessment (BDA). The Mamba architecture is designed to overcome the limitations of traditional Convolutional Neural Networks (CNNs) and Transformers, which struggle with capturing long-range dependencies and computational efficiency, respectively. The authors propose three specific frameworks, MambaBCD, MambaSCD, and MambaBDA, tailored for each CD task. These frameworks utilize the Visual Mamba architecture as the encoder to extract global spatial contextual information from multi-temporal images. For the change decoder, three spatio-temporal relationship modeling mechanisms are introduced to effectively capture spatio-temporal interactions in multi-temporal features. The proposed frameworks achieve competitive performance on five benchmark datasets without complex training strategies, demonstrating the potential of the Mamba architecture in CD tasks. The source code is available for public use.