TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
7x24hNews

Meituan Releases Trillion-Parameter Large Model LongCat-2.0, the First Trillion-Parameter Model to Complete Full-Process Training on a Domestic Computing Cluster

2026.06.30 - 02:57
Share

TechFlow News, June 30, according to Meituan's official announcement, Meituan officially launched the new generation large model LongCat-2.0 and open-sourced it simultaneously. The model totals 1.6T parameters and is the industry's first trillion-parameter model to complete full-process training and inference on a domestic computing cluster of 50,000 cards, natively supports 1M ultra-long context, with a core focus on code understanding, generation, and execution in Agentic Coding scenarios.

In terms of technology, LongCat-2.0 adopts the LongCat Sparse Attention (LSA) sparse attention mechanism, reducing long text computation complexity from quadratic to linear level; achieves token-level dynamic activation (33B~56B) through zero-computation expert mechanism; and introduces the MOPD architecture to integrate three sets of expert capabilities: Agent, Reasoning, and Interaction. In terms of training efficiency, the team spent three years overcoming domestic computing adaptation challenges, reducing the monthly average daily failure rate by over 70%, increasing training MFU by 1.5 times, with steady-state daily throughput exceeding 1T tokens/day.

In terms of performance evaluation, LongCat-2.0 scored 59.5 points on SWE-bench Pro, surpassing Gemini 3.1 Pro (54.2), GPT-5.5 (58.6), and Claude Opus 4.6 (57.3); scored 79.9 points on BrowseComp, reaching the level of frontier closed-source models.

Add to Favorites
Share to Social Media