The AI Museum
AI Analytics1,072 recordsRecounted 23 September 2026

Closed-weight models in 2022

Weightsopenclosedunknownfrontier only
Year20262025202420232022202120202019201820172016
2 filters active, showing 49 of 1,072 recordsClear all

49

Models

22 organisations

0%

Open weights

0 open, 49 closed

5

Frontier

By Epoch's flag

540B

Largest traceable

PaLM (540B)

2022

Span

26 with a traceable size

Releases

by month
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sept
Oct
Nov
Dec

Organisation

Country

Domain

Records

49
Closed-weight models in 2022, newest first. Each row links to the source Epoch AI records for it.
ModelOrganisationReleasedParametersWeights
CaLMUniversity of Oxford2022-1286M *Closed
DeepNashDeepMind2022-12—Closed
TranceptEveUniversity of Oxford2022-12—Closed
Vega v2Wuhan University2022-126BClosed
ALM 1.0Beijing Academy of Artificial Intelligence / BAAI2022-11335M *Closed
AR-LDMAlibaba2022-112BClosed
DiT-XL/2 + Discriminator GuidanceKorea Advanced Institute of Science and Technology (KAIST)2022-11—Closed
eDiff-INVIDIA2022-119B *Closed
Fusion in EncoderSamsung2022-11330M *Closed
Mogrifier RLSTM (WT2)DeepMind2022-1135MClosed
AlphaTensorDeepMind2022-10—Closed
Flan-PaLM 540BGoogle2022-10540BClosed
LMSI-PalmGoogle2022-10540BClosed
PhenakiUniversity College London (UCL)2022-102B *Closed
U-PaLM (540B)Google2022-10540BClosed
Make-A-VideoMeta AI2022-09—Closed
PaLIGoogle2022-0917B *Closed
SauTechSaudi Data and Artificial Intelligence Authority2022-09—Closed
AlexaTM 20BAmazon2022-0820BClosed
OmegaPLMMassachusetts Institute of Technology (MIT)2022-07670MClosed
CoCaGoogle Research2022-062BClosed
DITTOTsinghua University2022-06750MClosed
MetaLMMicrosoft Research2022-06—Closed
Minerva (540B)Google2022-06540BClosed
PartiGoogle Research2022-0620BClosed
DeBERTaV3large + KEARMicrosoft2022-05418MClosed
GatoDeepMind2022-051BClosed
ImagenGoogle Brain2022-058B *Closed
DALL·E 2OpenAI2022-044B *Closed
FlamingoDeepMind2022-0480BClosed
PaLM (540B)Google Research2022-04540BClosed
Sparse all-MLPMeta AI2022-049BClosed
XMC-GANGoogle Research2022-04—Closed
ChinchillaDeepMind2022-0370BClosed
DeepNetMicrosoft Research2022-033BClosed
GPT-3.5 (davinci-002)OpenAI2022-03—Closed
Make-A-SceneMeta AI2022-034B *Closed
MegaSynCollaborations Pharmaceuticals2022-03—Closed
Segatron-XL large, M=384 + HCPMicrosoft Research2022-03257MClosed
Statement Curriculum LearningOpenAI2022-03774M *Closed
AlphaCodeDeepMind2022-0241BClosed
FourCastNetNVIDIA2022-02—Closed
LaMDAGoogle2022-02137BClosed
Midjourney V1Midjourney2022-02—Closed
RETRO-7BDeepMind2022-028BClosed
ST-MoEGoogle2022-02269B *Closed
InstructGPT 1.3BOpenAI2022-011BClosed
InstructGPT 175BOpenAI2022-01175BClosed
InstructGPT 6BOpenAI2022-016BClosed

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