[Meituan Open-Sources LongCat-2.5 Model: 1.6 Trillion Total Parameters, Supports One Million Contexts] Jinshi Data, November 10th – Meituan officially open-sourced LongCat-2.5, using the MIT license, supporting commercial inference deployment and secondary training without additional licensing. LongCat-2.5 is a native multimodal model for long-range intelligent agent tasks, significantly improving end-to-end delivery capabilities in complex codebase development, complete project construction, visual coding, and professional office scenarios. The model has a total of 1.6 trillion parameters, with each token activating approximately 48 billion parameters, and supports one million contexts. Deep collaborative optimizations have also been made in model architecture, chip adaptation, and deployment strategies for domestically produced computing chips with limited GPU memory and bandwidth.