---
format: "aidr-story-markdown/v1"
id: "db313f233dd8bfe1c08d2991ec47156b79bfab26b66146ef952adfc8fe76535a"
canonical_url: "https://aidr.today/db313f23?lang=en"
title: "Meituan releases a 1.6-trillion-parameter agent model that never touched an Nvidia GPU"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-15T10:06:06.000Z"
category: "Models"
topics: ["agent"]
source_urls: ["https://the-decoder.com/meituan-releases-a-1-6-trillion-parameter-agent-model-that-never-touched-an-nvidia-gpu/"]
summary: "Meituan released the weights for LongCat-2.5, a 1.6-trillion-parameter model built for AI agents. The model was trained entirely on Chinese hardware and runs on 194 Huawei Ascend chips. On Terminal Bench, it scores 88.4 points."
---

# Meituan releases a 1\.6\-trillion\-parameter agent model that never touched an Nvidia GPU

> [Open the canonical story](<https://aidr.today/db313f23?lang=en>)

**Published:** 2026-11-15T10:06:06.000Z
**Category:** Models
**Topics:** agent

## Summary

Meituan released the weights for LongCat\-2\.5, a 1\.6\-trillion\-parameter model built for AI agents\. The model was trained entirely on Chinese hardware and runs on 194 Huawei Ascend chips\. On Terminal Bench, it scores 88\.4 points\.

## Sources

- [Story source](<https://the-decoder.com/meituan-releases-a-1-6-trillion-parameter-agent-model-that-never-touched-an-nvidia-gpu/>)