Event ·Official first-hand

Meituan LongCat-2.0：1.6万亿参数模型完全基于国产芯片训练

美团发布LongCat-2.0，一个参数量达1.6万亿的AI模型，所有训练均使用国产AI芯片完成。该模型利用美团外卖配送数据驱动，表明生活服务场景也能支撑前沿大模型研究。模型规模超过OpenAI GPT-3的1750亿参数约9倍，但训练成本因国产芯片而受到关注。LongCat-2.0现已在内部用于配送调度等业务，未公开计划开源。

Current read

美团用外卖数据训出1.6万亿参数大模型，而且全用国产芯片，技术路线和落地场景都很有意思。

11 sources68° AI heatLast updated 2026/7/8 08:08:23

Evidence

11 sources

Primary sourcepandaily

2026/7/8 08:08:23View source

Related source 1Taranjeet

2026/7/9 18:00:07View source

Related source 2AI Will

2026/7/10 04:21:47View source

Related source 3The Rundown AI

2026/7/6 10:30:14View source

Related source 4小互

2026/7/6 14:19:37View source

Related source 5@OpenAIDevs

2026/7/6 22:13:21View source

Related source 6marktechpost

2026/7/7 04:35:09View source

Related source 7@koltregaskes

2026/7/7 08:48:29View source

Related source 8IT之家

2026/7/7 08:58:34View source

Related source 9lmarena.ai

2026/7/7 19:51:02View source

Related source 10OpenAI Blog

2026/7/8 00:00:00View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story