Saved in:
Bibliographic Details
Main Authors: ai, YuanLab., :, Wu, Shawn, Luo, Jiangang, Chen, Darcy, Wang, Sean, Li, Louie, Wang, Allen, Zhao, Xudong, Yu, Tong, Li, Bach, Shen, Joseph, Ma, Gawain, Jia, Jasper, Mao, Marcus, Wang, Claire, He, Hunter, Wang, Carol, Zhang, Zera, Wang, Jason, Shen, Chonly, Zhang, Leo, Chen, Logan, Meng, Qasim, Gong, James, Zhao, Daniel, Zheng, Penn, Zhu, Owen
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.14327
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914369274642432
author ai, YuanLab.
:
Wu, Shawn
Luo, Jiangang
Chen, Darcy
Wang, Sean
Li, Louie
Wang, Allen
Zhao, Xudong
Yu, Tong
Li, Bach
Shen, Joseph
Ma, Gawain
Jia, Jasper
Mao, Marcus
Wang, Claire
He, Hunter
Wang, Carol
Zhang, Zera
Wang, Jason
Shen, Chonly
Zhang, Leo
Chen, Logan
Meng, Qasim
Gong, James
Zhao, Daniel
Zheng, Penn
Zhu, Owen
author_facet ai, YuanLab.
:
Wu, Shawn
Luo, Jiangang
Chen, Darcy
Wang, Sean
Li, Louie
Wang, Allen
Zhao, Xudong
Yu, Tong
Li, Bach
Shen, Joseph
Ma, Gawain
Jia, Jasper
Mao, Marcus
Wang, Claire
He, Hunter
Wang, Carol
Zhang, Zera
Wang, Jason
Shen, Chonly
Zhang, Leo
Chen, Logan
Meng, Qasim
Gong, James
Zhao, Daniel
Zheng, Penn
Zhu, Owen
contents We introduce Yuan3.0 Ultra, an open-source Mixture-of-Experts (MoE) large language model featuring 68.8B activated parameters and 1010B total parameters, specially designed to enhance performance on enterprise scenarios tasks while maintaining competitive capabilities on general purpose tasks. We propose Layer-Adaptive Expert Pruning (LAEP) algorithm designed for the pre-training stage of MoE LLMs. In contrast to previous expert pruning approaches that operate primarily in the post-training phase, the proposed algorithm enhances training efficiency by selectively pruning underutilized experts and reorganizing experts across computing devices according to token distribution statistics. Comprehensive experiments demonstrate that LAEP effectively reduces model size and substantially improves pre-training efficiency. When pre-training Yuan3.0 Ultra from scratch original with 1515B parameters, this algorithm delivers a 49\% boost in pre-training efficiency and a 33.3\% reduction in total parameters, while preserving the model's outstanding multi-domain performance. On enterprise scenario benchmarks including Docmatix, ChatRAG, SummEval and MMTab, Yuan3.0 Ultra achieves leading accuracy. The model and codes are publicly available at https://github.com/Yuan-lab-LLM/Yuan3.0-Ultra.
format Preprint
id arxiv_https___arxiv_org_abs_2601_14327
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Yuan3.0 Ultra: A Trillion-Parameter Enterprise-Oriented MoE LLM
ai, YuanLab.
:
Wu, Shawn
Luo, Jiangang
Chen, Darcy
Wang, Sean
Li, Louie
Wang, Allen
Zhao, Xudong
Yu, Tong
Li, Bach
Shen, Joseph
Ma, Gawain
Jia, Jasper
Mao, Marcus
Wang, Claire
He, Hunter
Wang, Carol
Zhang, Zera
Wang, Jason
Shen, Chonly
Zhang, Leo
Chen, Logan
Meng, Qasim
Gong, James
Zhao, Daniel
Zheng, Penn
Zhu, Owen
Machine Learning
Artificial Intelligence
Computation and Language
We introduce Yuan3.0 Ultra, an open-source Mixture-of-Experts (MoE) large language model featuring 68.8B activated parameters and 1010B total parameters, specially designed to enhance performance on enterprise scenarios tasks while maintaining competitive capabilities on general purpose tasks. We propose Layer-Adaptive Expert Pruning (LAEP) algorithm designed for the pre-training stage of MoE LLMs. In contrast to previous expert pruning approaches that operate primarily in the post-training phase, the proposed algorithm enhances training efficiency by selectively pruning underutilized experts and reorganizing experts across computing devices according to token distribution statistics. Comprehensive experiments demonstrate that LAEP effectively reduces model size and substantially improves pre-training efficiency. When pre-training Yuan3.0 Ultra from scratch original with 1515B parameters, this algorithm delivers a 49\% boost in pre-training efficiency and a 33.3\% reduction in total parameters, while preserving the model's outstanding multi-domain performance. On enterprise scenario benchmarks including Docmatix, ChatRAG, SummEval and MMTab, Yuan3.0 Ultra achieves leading accuracy. The model and codes are publicly available at https://github.com/Yuan-lab-LLM/Yuan3.0-Ultra.
title Yuan3.0 Ultra: A Trillion-Parameter Enterprise-Oriented MoE LLM
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2601.14327