LongCat-2.5 is Meituan LongCat's latest natively multimodal agentic model with 1.6T total parameters and ~48B active parameters. Built for long-horizon agentic tasks, it features stronger visual understanding and agentic coding capabilities for complex coding projects, creative design, and 3D tasks. The API is available at longcat.ai/platform with 10 million tokens provided on signup.
LongCat-2.5 is Meituan’s native multimodal mixture-of-experts model for long-horizon agent tasks. Meituan reports 1.6 trillion total parameters and about 48 billion activated per token. It builds on LongCat-2.0, combining visual understanding with agentic coding for complex codebases and full-project builds, as well as creative design and 3D tasks. The model processes visual and text inputs through a shared token-based representation. Its official API documentation describes image and video input, including image description, screenshot text reading, chart analysis, and video understanding.
