ModelRadar
/

LongCat-2.5

Meituan LongCat🇨🇳 CNMajor lab· Nov 4, 2026

TextVisionopen weights

LongCat-2.5 is a native multimodal MoE model with 1.6 trillion total parameters and about 48 billion active per token, built for long-horizon agentic work. It pairs visual understanding with agentic coding for codebase development, full-project builds and 3D tasks.

Summary of the model card

License
mit
Modalities
text, image → text

2 variants

W4A4HF
StandardHF

Sources

HuggingFacePaper
Deutsch

LongCat-2.5 ist ein nativ multimodales MoE-Modell mit 1,6 Billionen Gesamtparametern und etwa 48 Milliarden aktiven pro Token für agentische Aufgaben. Es verbindet Bildverständnis mit agentischem Coding für Codebasen, komplette Projektbuilds und 3D-Aufgaben.

Provenance

Detected
Nov 9, 2026
First source
huggingface
Description
Summary of the model card
Editorially reviewed
—