The AI Museum
AI Analytics1,072 recordsRecounted 23 September 2026

Open-weight vision models

Weightsopenclosedunknownfrontier only
Year20262025202420232022202120202019201820172016
2 filters active, showing 57 of 1,072 recordsClear all

57

Models

37 organisations

100%

Open weights

57 open, 0 closed

1

Frontier

By Epoch's flag

72B

Largest traceable

NVLM-D 72B

2014–2026

Span

31 with a traceable size

Releases

by year
2014201520162017201820192020202120222023202420252026

Organisation

Country

Domain

Records

newest 50 of 57
Open-weight vision models, newest first. Each row links to the source Epoch AI records for it.
ModelOrganisationReleasedParametersWeights
MolmoAct 2Allen Institute for AI2026-045B *Open
EXAONE Path 2.0LG AI Research2025-07175MOpen
Eagle 2NVIDIA2025-019BOpen
NVILA 15BNVIDIA2024-1215BOpen
NVLM-D 72BNVIDIA2024-1072BOpen
CogAgentTsinghua University2023-1218B *Open
mPLUG-Owl2Alibaba2023-117B *Open
SPHINX (Llama 2 13B)Shanghai AI Lab2023-1120B *Open
DINOv2Facebook AI Research2023-041BOpen
Segment Anything ModelMeta AI2023-04636MOpen
SigLIP 400MGoogle DeepMind2023-03400MOpen
BLIP-2 (Q-Former)Salesforce Research2023-011BOpen
EVA-01Beijing Academy of Artificial Intelligence / BAAI2022-111BOpen
InternImageShanghai AI Lab2022-111BOpen
ViT-G (model soup)University of Washington2022-032BOpen
data2vec (vision)Meta AI2022-01705M *Open
DeticMeta AI2022-0188M *Open
Masked Autoencoders ViT-HFacebook AI Research2021-11632M *Open
Swin Transformer V2 (SwinV2-G)Microsoft Research Asia2021-113BOpen
TrOCRBeihang University2021-09558MOpen
YOLOX-XMegvii Inc2021-0899M *Open
SEERFacebook AI Research2021-071BOpen
Denoising Diffusion Probabilistic Models (LSUN Bedroom)University of California (UC) Berkeley2021-06256MOpen
EfficientNetV2-XLGoogle2021-06208MOpen
Transformer local-attention (NesT-B)Google Cloud2021-0590MOpen
DeiT-BMeta AI2021-0186MOpen
ViT-Base/32Google Brain2020-1086M *Open
ViT-Huge/14Google Brain2020-10632MOpen
EfficientDetGoogle Brain2020-0777M *Open
DETRFacebook2020-0560MOpen
Once for AllMIT-IBM Watson AI Lab2020-048MOpen
SimCLRGoogle Brain2020-02375M *Open
StarGAN v2NAVER2019-12—Open
MoCoFacebook AI2019-11375M *Open
BigBiGANGoogle2019-0786M *Open
FixRes ResNeXt-101 WSLFacebook AI2019-06829M *Open
LaNet-L (CIFAR-10)Brown University2019-0644MOpen
EfficientNet-L2Google2019-05480M *Open
ResNeXt-101 Billion-scaleFacebook AI2019-05193M *Open
Big-Little NetIBM2018-0777M *Open
ResNeXt-101 32x48dFacebook2018-05829MOpen
PyramidNetKorea Advanced Institute of Science and Technology (KAIST)2017-0926M *Open
3DMM-CNNUniversity of Southern California2016-1245MOpen
EnhanceNetMax Planck Institute for Intelligent Systems2016-12814,464Open
HR-ResNet101Carnegie Mellon University (CMU)2016-1245MOpen
YOLOv2University of Washington2016-1251M *Open
DLDL (PASCAL)University of Oxford2016-11564M *Open
ResNeXt-101 (64×4d)University of California San Diego2016-1183MOpen
ResNeXt-50University of California San Diego2016-1125M *Open
DenseNet-264Tsinghua University2016-0834MOpen

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