# Meituan Launches LongCat-2.5 Multimodal Agentic Model

> Meituan released LongCat-2.5, a 1.6-trillion-parameter model with a one-million-token context window aimed at coding and GUI tasks, part of a wider push by Chinese labs toward long-running agent workflows.

- Type: AI news
- Page: https://completeaitraining.com/news/meituan-launches-longcat-25-multimodal-agentic-model/
- Published: 2026-11-02
- Updated: 2026-11-03
- Author: Jeroen Erne
- Jobs: Creatives, IT and development, Marketing, Operations, Product development
- Reading time: 1 min

Meituan has introduced LongCat-2.5, a mixture-of-experts model with 1.6 trillion parameters and a one-million-token context window. The model is positioned for coding and GUI (graphical user interface) tasks, meaning work that involves both generating code and interacting with software interfaces.

The release reflects a broader pattern among Chinese labs, which are combining large context windows, multimodal interaction and domestic accelerators in an effort to compete in long-horizon agent workflows. Long-horizon workflows generally involve agents carrying out multi-step tasks over extended runs rather than single prompts, which places demands on how much context a model can hold and how it handles different input types.

For teams working in development, product and operations, the stated focus areas of coding and GUI tasks point to agent use cases that touch both code generation and direct interaction with application interfaces. The summary does not detail availability, pricing, benchmark results or which accelerators are used.

---

This is the Markdown form of https://completeaitraining.com/news/meituan-launches-longcat-25-multimodal-agentic-model/. The whole site is mapped at https://completeaitraining.com/llms.txt.