Meituan's LongCat team explains why they chose LongCat: Behind the scenes of opening-source LongCat-2.5 and training on domestic computing power
Automated summary · Based on collected summaries; not individually verified
Prepared by HotAI
What happened
Meituan has officially released and opened-source its native multimodal model LongCat-2.5, designed for long-horizon agent tasks. The model was fully trained and run on domestic computing hardware, building upon LongCat-2.0 to enhance native visual understanding and agent programming capabilities.
Background
What is a multimodal model? Checking image, audio, and text capabilities separately
Multimodal models process or relate multiple forms such as text and images, potentially audio or video. Input support does not imply corresponding generation, and separate modality support does not guarantee complex joint tasks. Check model and interface limits and parsing; uploads alone do not establish accurate understanding.
What is an AI agent? Compare it with a chatbot through a document task
This article focuses on LLM-powered AI agents: systems that choose an action for a goal, obtain external results through tools, and use those results to continue or stop. An agent organizing meeting records might search files, discover that a date is missing, and inspect another record. A basic Q&A interaction generates a response; an agent also dynamically directs execution. One tool call alone does not demonstrate that control. Check the tool records and final result.
Reading notes
The event summary comes from collected excerpts. Background passages quote existing HotAI articles to explain terms and do not establish new facts about this event. Check the original sources for product results, conditions and subsequent updates.
Original source
Sources and collection timeline
| Source | Report title | Original publication time | Collected by HotAI |
|---|---|---|---|
| mp.weixin.qq.com | Meituan's LongCat team explains why they chose LongCat: Behind the scenes of opening-source LongCat-2.5 and training on domestic computing power | 2026-11-12 20:18 | 2026-11-12 20:37 |