Meituan LongCat releases flagship model LongCat-2.0
Original title:The full model behind "Owl Alpha" on @OpenRouter is here🦉 Let's meet @Meituan_LongCat 's latest fla...
AI summary
Meituan LongCat has launched its flagship model LongCat-2.0, featuring a 1.6T-parameter MoE architecture (about 48B active parameters) with native support for a 1M context window. Pricing is $0.015/1M tokens for input cache, $0.75/1M tokens for input, and $2.95/1M tokens for output. The model is designed for Agentic Coding and includes three key technologies: LSA sparse attention for efficient 1M scaling; Zero-Compute Experts, which dynamically activate 33B–56B parameters per token with no wasted compute; and MOPD, which divides experts into Agent, Reasoning, and Interaction groups with task-gated routing. It scores 59.5 on SWE-bench Pro, approaching mainstream closed-source models. It is now available on SiliconFlow Day 0.
Key Takeaways
- Meituan LongCat has launched its flagship model LongCat-2.
- 0,featuring a 1.
- 6T-parameter MoE architecture (about 48B active parameters) with native support for a 1M context…
- Pricing is $0.
- 015/1M tokens for input cache,$0.
Quick Facts
- Published
- Updated
- Category
- Model Releases
- AI score
- 67
美团龙猫的 LongCat-2.0 专为 agentic coding 设计的 MoE 模型,架构上三种专家分工有点意思,SWE-bench 59.5 接近闭源水平,已经能在硅基流动上直接调,做 coding agent 的可以跑跑看。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Meituan LongCat releases flagship model LongCat-2.0. June 30, 2026. https://new.nat6.net/en/news/cmr08dev204qjslkit52gdrur/