Closed-weight models from DeepMind
32
Models
1 organisations
0%
Open weights
0 open, 32 closed
5
Frontier
By Epoch's flag
280B
Largest traceable
Gopher (280B)
2015–2023
Span
14 with a traceable size
Organisation
Country
Records
32| Model | Organisation | Released | Parameters | Weights |
|---|---|---|---|---|
| DreamerV3 | DeepMind | 2023-01 | 200M * | Closed |
| DeepNash | DeepMind | 2022-12 | — | Closed |
| Mogrifier RLSTM (WT2) | DeepMind | 2022-11 | 35M | Closed |
| AlphaTensor | DeepMind | 2022-10 | — | Closed |
| Gato | DeepMind | 2022-05 | 1B | Closed |
| Flamingo | DeepMind | 2022-04 | 80B | Closed |
| Chinchilla | DeepMind | 2022-03 | 70B | Closed |
| AlphaCode | DeepMind | 2022-02 | 41B | Closed |
| RETRO-7B | DeepMind | 2022-02 | 8B | Closed |
| Gopher (280B) | DeepMind | 2021-12 | 280B | Closed |
| Student of Games | DeepMind | 2021-12 | — | Closed |
| GOAT | DeepMind | 2021-07 | 3M * | Closed |
| Agent57 | DeepMind | 2020-03 | — | Closed |
| Perceiver IO (optical flow) | DeepMind | 2020-02 | 28M | Closed |
| AlphaFold | DeepMind | 2020-01 | 16M * | Closed |
| MuZero | DeepMind | 2019-11 | 37M * | Closed |
| AlphaStar | DeepMind | 2019-10 | 139M | Closed |
| Mogrifier (d2, MoS2, MC) + dynamic eval | DeepMind | 2019-09 | 35M | Closed |
| CPC v2 | DeepMind | 2019-05 | 303M * | Closed |
| Hanabi 4 player | DeepMind | 2019-02 | 764,000 | Closed |
| FTW (For The Win) | DeepMind | 2018-07 | 126M * | Closed |
| Relational Memory Core | DeepMind | 2018-06 | — | Closed |
| IMPALA | DeepMind | 2018-02 | 2M | Closed |
| AlphaZero | DeepMind | 2017-12 | — | Closed |
| VQ-VAE | DeepMind | 2017-11 | — | Closed |
| AlphaGo Master | DeepMind | 2017-10 | — | Closed |
| AlphaGo Zero | DeepMind | 2017-10 | 46M | Closed |
| Rainbow DQN | DeepMind | 2017-10 | — | Closed |
| AWD-LSTM | DeepMind | 2017-07 | 24M | Closed |
| NoisyNet-Dueling | DeepMind | 2017-06 | — | Closed |
| AlphaGo Lee | DeepMind | 2016-01 | — | Closed |
| AlphaGo Fan | DeepMind | 2015-10 | 8M * | 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.