The AI Museum
AI Analytics1,092 recordsRecounted 5 November 2026

Closed-weight models in 2023

Weightsopenclosedunknownfrontier only
Year20262025202420232022202120202019201820172016
2 filters active, showing 60 of 1,092 recordsClear all

60

Models

29 organisations

0%

Open weights

0 open, 60 closed

8

Frontier

By Epoch's flag

1.1T

Largest traceable

PanGu-Σ

2023

Span

21 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 60
Closed-weight models in 2023, newest first. Each row links to the source Epoch AI records for it.
ModelOrganisationReleasedParametersWeights
CoReTsinghua University2023-1212B *Closed
Gemini 1.0 ProGoogle DeepMind2023-12—Closed
Gemini 1.0 UltraGoogle DeepMind2023-12—Closed
Gemini Nano-1Google DeepMind2023-122BClosed
Gemini Nano-2Google DeepMind2023-123BClosed
GQA-8-XXLGoogle Research2023-1211BClosed
Mamba-24M (SC09)Carnegie Mellon University (CMU)2023-1223MClosed
W.A.L.TStanford University2023-125BClosed
AndesGPTOppo Mobile Telecommunications2023-11—Closed
Claude 2.1Anthropic2023-11—Closed
Cohere EmbedCohere2023-11—Closed
GNoME for crystal discoveryGoogle DeepMind2023-1116M *Closed
GPT-4 Turbo (Nov 2023)OpenAI2023-11—Closed
Inflection-2Inflection AI2023-11—Closed
OmniVecTensorTour2023-11—Closed
PPLX-70B-OnlinePerplexity2023-1170B *Closed
RoFormerZhuiyi Technology2023-11110MClosed
RT-TrajectoryGoogle DeepMind2023-11—Closed
CODEFUSION (Python)Microsoft2023-1075MClosed
DALL·E 3OpenAI2023-10—Closed
DiT-XL/2 + CADSETH Zurich2023-10675M *Closed
ERNIE 4.0Baidu2023-10—Closed
PaLI-3Google DeepMind2023-105BClosed
RT-2-XGoogle DeepMind2023-1055BClosed
AlphaMissenseGoogle DeepMind2023-0993M *Closed
Amazon TitanAmazon2023-09200B *Closed
GPT-3.5 Turbo InstructOpenAI2023-0920B *Closed
GPT-4VOpenAI2023-09—Closed
Robot ParkourShanghai Qi Zhi institute2023-09500,000Closed
GGNNWestlake University2023-08—Closed
SwiftIntel Labs2023-0856,804 *Closed
AudioLMGoogle Research2023-072B *Closed
Claude 2Anthropic2023-07—Closed
GPT3-2.7B (FlashAttention-2)Stanford University2023-073BClosed
InternLMShanghai AI Lab2023-07104BClosed
RT-2Google DeepMind2023-0755BClosed
xTrimoPGLM -100BTsinghua University2023-07100BClosed
ERNIE 3.5Baidu2023-06—Closed
GPT-3.5 TurboOpenAI2023-0620B *Closed
GPT-4 (Jun 2023)OpenAI2023-061.8T *Closed
LTM-1Magic2023-06—Closed
DPO on Pythia-2.8BStanford University2023-053BClosed
Med-PaLM 2Google Research2023-05340B *Closed
PaLI-XGoogle Research2023-0555B *Closed
PaLM 2Google2023-05340B *Closed
Agile Soccer RobotGoogle DeepMind2023-04—Closed
BloombergGPTBloomberg2023-0351BClosed
ClaudeAnthropic2023-03—Closed
FireflyAdobe2023-03—Closed
Gen-2Runway2023-03—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