Meituan launches LongCat-2.5, a native multimodal model for long-horizon agents

Meituan's 1.6-trillion-parameter model adds native vision to long-horizon coding, with a 1M-token API window and GitHub and Hugging Face releases still to come.

By · Published · Updated

Primary source: Hacker News

Why it matters

LongCat-2.5 shows a large platform company pushing frontier-scale agentic coding with native vision, a 1M-token window, and low cached-input pricing.

A yellow four-wheeled robot marked 美团 stands beside a matching paper bag, with floating screens showing pedestrians crossing streets.

Meituan, co-founded by Wang Xing, released LongCat-2.5 on November 2nd, according to Meituan's announcement. The native multimodal model is built for long-horizon agentic work, with 1.6 trillion total parameters and approximately 48 billion activated per token. The release pushes Meituan's model line further into sustained coding work, where an agent has to hold a project together across many steps.…

When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Reader comments

Conversation for this story loads after sign-in.