The AI Museum
AI Analytics1,072 recordsRecounted 23 September 2026

Closed-weight language models from China

Weightsopenclosedunknownfrontier only
Year20262025202420232022202120202019201820172016
3 filters active, showing 45 of 1,072 recordsClear all

45

Models

26 organisations

0%

Open weights

0 open, 45 closed

3

Frontier

By Epoch's flag

2.4T

Largest traceable

Qwen 3.8 Max

2015–2026

Span

18 with a traceable size

Releases

by year
201520162017201820192020202120222023202420252026

Organisation

Country

Domain

Records

45
Closed-weight language models from China, newest first. Each row links to the source Epoch AI records for it.
ModelOrganisationReleasedParametersWeights
Step 5 PreviewStepFun2026-09600B *Closed
GLM-5.3Z.ai (Zhipu AI)2026-08744BClosed
Qwen 3.8 MaxAlibaba2026-072.4TClosed
Qwen3.7-PlusAlibaba2026-06—Closed
Qwen 3.6 PlusAlibaba2026-04—Closed
MiMo-V2-ProXiaomi Corp2026-031.0TClosed
Qwen3.5-Omni-FlashAlibaba2026-03—Closed
Qwen3.5-Omni-PlusAlibaba2026-03—Closed
Qwen3-Max-ThinkingAlibaba2026-01—Closed
Qwen3-MaxAlibaba2025-091.0T *Closed
Pangu UltraHuawei2025-04135BClosed
Hunyuan-TurboSTencent2025-03560BClosed
Doubao-1.5-proByteDance2025-01—Closed
Doubao-proByteDance2024-10500B *Closed
Yi-Lightning01.AI2024-10—Closed
Hunyuan TurboTencent2024-09—Closed
GLM-4-PlusZ.ai (Zhipu AI)2024-08—Closed
HairuoInspur2024-08—Closed
GLM-4 (0520)Z.ai (Zhipu AI)2024-05—Closed
Yi-Large01.AI2024-05100B *Closed
MegaScale (Production)ByteDance2024-02530B *Closed
Kimi ExplorerMoonshot2024-01—Closed
AndesGPTOppo Mobile Telecommunications2023-11—Closed
RoFormerZhuiyi Technology2023-11110MClosed
InternLMShanghai AI Lab2023-07104BClosed
ERNIE 3.5Baidu2023-06—Closed
PanGu-ΣHuawei Noah's Ark Lab2023-031.1TClosed
Vega v2Wuhan University2022-126BClosed
ALM 1.0Beijing Academy of Artificial Intelligence / BAAI2022-11335M *Closed
DITTOTsinghua University2022-06750MClosed
ERNIE 3.0 TitanBaidu2021-12260BClosed
Yuan 1.0Inspur2021-10246BClosed
PermuteFormerPeking University2021-09150M *Closed
Adaptive Input Transformer + RDMicrosoft Research Asia2021-06247M *Closed
PLUGAlibaba2021-0427BClosed
KEPLERTsinghua University2020-11125MClosed
NAS+ESS (23M)Northeastern University (China)2020-0523MClosed
TCAN (WT2)Nanjing University2020-0233MClosed
MMLSTM (PTB)Beijing University of Posts and Telecommunications2019-1221M *Closed
MMLSTM (WT-2)Beijing University of Posts and Telecommunications2019-1232M *Closed
EN^2AS with performance rewardBeijing Institute of Technology2019-0723M *Closed
True-Regularization+Finetune+Dynamic-EvalMobvoi2019-047MClosed
AWD-LSTM-MoS + dynamic evaluation (WT2, 2018)Peking University2018-0935MClosed
DL scaling LMBaidu2017-12177M *Closed
genCNN + dyn evalChinese Academy of Sciences2015-038M *Closed

An asterisk marks a parameter figure the source does not consider traceable to the people who built the model. That judgement is the source’s and it has false negatives: Kimi K3 is marked speculative at 2.8T while Moonshot’s own model card states the figure. Why that distinction carries the whole exhibit.

Counts are of models Epoch AI records as notable, not of every model released. The most recent month is always incomplete, because a model enters the dataset when it is reviewed rather than when it ships.

Epoch AI, 'Data on AI Models'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/data/ai-models-documentation' Used under CC BY 4.0.

How this museum is made