Report

Meituan's LongCat-2.5 reportedly arrives as a ~1.6T-parameter multimodal agent model

LongCat-2.5 is said to have ~48B active parameters and outperform Opus 5 and GPT-5.6-sol on AutomationBench and BrowserComp.

2 Sources, 11h ago, first seen 11h ago

TLDR

LongCat-2.5 is described as a mixture-of-experts model built for long-horizon agents, software engineering and multimodal tasks. A post claims it outperforms Opus 5 and GPT-5.6-sol on AutomationBench and BrowserComp. The listed prices per million tokens are ¥2 for uncached input, ¥0.04 for cached input and ¥8 for output.

Today's Rank

—

Not ranked yet

Combined views

—

2 Sources, first seen 11h ago