Meituan has open-sourced the LongCat-2.5 agent model with 16 trillion parameters, noting that Nvidia GPUs were not used during training.
Automated summary · Based on collected summaries; not individually verified
Prepared by HotAI
What happened
Meituan has opened up LongCat-2.5, a 16 trillion-parameter MoE model designed for agents. It activates about 48 billion parameters per token and uses an MIT license. The official deployment solution relies on 194 Huawei Ascend accelerators, with no Nvidia GPUs used throughout.
Background
What is a token? Its relationship to text length, usage, and context budgets
A token is a discrete unit used to process text, potentially a character, part of a word, or punctuation depending on the tokenizer. It is not a fixed number of Chinese characters or English words. Budgets and usage often use tokens, but accounting differs by model and interface. Use the target model’s counting method and inspect reported usage.
What is a GPU? Why graphics processors also run AI computation
A GPU performs highly parallel computation and is used beyond rendering in machine learning. It is not a model, and owning one does not make every model runnable. Workload, memory and software matter. Detection, loading and correct task completion require separate checks.
Reading notes
The event summary comes from collected excerpts. Background passages quote existing HotAI articles to explain terms and do not establish new facts about this event. Check the original sources for product results, conditions and subsequent updates.
Original source
Sources and collection timeline
| Source | Report title | Original publication time | Collected by HotAI |
|---|---|---|---|
| The Decoder | Meituan has open-sourced the LongCat-2.5 agent model with 16 trillion parameters, noting that Nvidia GPUs were not used during training. | 2026-11-15 18:06 | 2026-11-15 18:47 |