The AI Museum
AI Analytics1,092 recordsRecounted 5 November 2026

Models in 2017

Weightsopenclosedunknownfrontier only
Year20262025202420232022202120202019201820172016
1 filter active, showing 55 of 1,092 recordsClear all

55

Models

35 organisations

4%

Open weights

1 open, 25 closed

3

Frontier

By Epoch's flag

9B

Largest traceable

MoE-Multi

2017

Span

13 with a traceable size

Releases

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

Organisation

Country

Domain

Records

newest 50 of 55
Models in 2017, newest first. Each row links to the source Epoch AI records for it.
ModelOrganisationReleasedParametersWeights
AlphaZeroDeepMind2017-12—Closed
DL scaling ImageBaidu2017-12121MClosed
DL scaling LMBaidu2017-12177M *Closed
DL scaling speechBaidu2017-12193M *Closed
AWD-LSTM-MoS + dynamic evaluation (WT2, 2017)Carnegie Mellon University (CMU)2017-1135M *Closed
TriNetVisual Computing Institute2017-11—Not recorded
VQ-VAEDeepMind2017-11—Closed
AlphaGo MasterDeepMind2017-10—Closed
AlphaGo ZeroDeepMind2017-1046MClosed
DCN+Salesforce Research2017-10—Closed
Fraternal dropout + AWD-LSTM 3-layer (WT2)Jagiellonian University2017-1034M *Closed
PhraseCondCarnegie Mellon University (CMU)2017-10—Not recorded
ProgressiveGANNvidia2017-10—Not recorded
Rainbow DQNDeepMind2017-10—Closed
S-NormUniversity of Washington2017-10—Not recorded
AWD-LSTM+WT+Cache+IOG (WT2)NTT Communication Science Laboratories2017-0953MClosed
ISSDuke University2017-0911MClosed
LSTM + dynamic evalUniversity of Edinburgh2017-0950M *Closed
PyramidNetKorea Advanced Institute of Science and Technology (KAIST)2017-0926M *Open
SENet (ImageNet)Chinese Academy of Sciences2017-0928M *Not recorded
AWD-LSTM - 3-layer LSTM (tied) + continuous cache pointer (WT2)Salesforce Research2017-0833MClosed
EI-REHN-1000DKorea Advanced Institute of Science and Technology (KAIST)2017-0819MClosed
GL-LWGC-AWD-MoS-LSTM + dynamic evaluation (WT2)Ben-Gurion University of the Negev2017-0838MClosed
LibratusCarnegie Mellon University (CMU)2017-08—Closed
NeuMF (Pinterest)Shandong University2017-08—Not recorded
OpenAI TI7 DOTA 1v1OpenAI2017-08—Not recorded
RetinaNet-R101Facebook AI Research2017-0853MNot recorded
RetinaNet-R50Facebook AI Research2017-0834M *Not recorded
AWD-LSTMDeepMind2017-0724MClosed
ConvS2S (ensemble of 8 models)Meta AI2017-07—Not recorded
DeepLocTechnical University of Denmark2017-07—Closed
GSMPeking University2017-07—Not recorded
JFTGoogle Research2017-0745MNot recorded
NASNet-AGoogle Brain2017-0789M *Not recorded
PSPNetChinese University of Hong Kong (CUHK)2017-07—Not recorded
ShuffleNet v1Megvii Inc2017-072M *Not recorded
DeepLabV3Google2017-06—Not recorded
EDSRSeoul National University2017-06—Not recorded
HRAMaluuba2017-06—Not recorded
Inflated 3D ConvNetDeepMind2017-06—Not recorded
NoisyNet-DuelingDeepMind2017-06—Closed
PointNet++Stanford University2017-06—Not recorded
Reading Twice for NLUDeepMind2017-06—Not recorded
TransformerGoogle Research2017-06213MClosed
Mnemonic ReaderFudan University2017-05—Not recorded
SRGANTwitter2017-05—Closed
DeepLab (2017)Johns Hopkins University2017-04—Not recorded
MobileNetGoogle2017-044M *Not recorded
Mask R-CNNFacebook AI Research2017-03—Not recorded
Prototypical networksUniversity of Toronto2017-03—Not recorded

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