China AI DataTokenHeart News ↗
← Back to dashboardDATA DETAIL / MODELS

Model release pace

Monthly counts by original publication date. Completed months compare with their previous month; the current month compares equal completed days. A zero denominator has no percentage growth. This is not an exhaustive model census.

Model release pace

Model release pace2024-10-01: 11.00; 2024-11-01: 8.00; 2024-12-01: 13.00; 2025-01-01: 10.00; 2025-02-01: 4.00; 2025-03-01: 10.00; 2025-04-01: 6.00; 2025-05-01: 7.00; 2025-06-01: 4.00; 2025-07-01: 10.00; 2025-08-01: 11.00; 2025-09-01: 8.00; 2025-10-01: 6.00; 2025-11-01: 19.00; 2025-12-01: 12.00; 2026-01-01: 1.00; 2026-02-01: 9.00; 2026-03-01: 9.00; 2026-04-01: 17.00; 2026-05-01: 6.00; 2026-06-01: 9.00; 2026-07-01: 7.00; 2026-08-01: 8.00; 2026-09-01: 12.0009.5192024-10-01: 11.00 2024-11-01: 8.00 2024-12-01: 13.00 2025-01-01: 10.00 2025-02-01: 4.00 2025-03-01: 10.00 2025-04-01: 6.00 2025-05-01: 7.00 2025-06-01: 4.00 2025-07-01: 10.00 2025-08-01: 11.00 2025-09-01: 8.00 2025-10-01: 6.00 2025-11-01: 19.00 2025-12-01: 12.00 2026-01-01: 1.00 2026-02-01: 9.00 2026-03-01: 9.00 2026-04-01: 17.00 2026-05-01: 6.00 2026-06-01: 9.00 2026-07-01: 7.00 2026-08-01: 8.00 2026-09-01: 12.00 10/0109/01 UTC

Monthly release growth

Monthly release growth2024-10-01: -15.38%; 2024-11-01: -27.27%; 2024-12-01: 62.50%; 2025-01-01: -23.08%; 2025-02-01: -60.00%; 2025-03-01: 150.00%; 2025-04-01: -40.00%; 2025-05-01: 16.67%; 2025-06-01: -42.86%; 2025-07-01: 150.00%; 2025-08-01: 10.00%; 2025-09-01: -27.27%; 2025-10-01: -25.00%; 2025-11-01: 216.67%; 2025-12-01: -36.84%; 2026-01-01: -91.67%; 2026-02-01: 800.00%; 2026-03-01: 0.00%; 2026-04-01: 88.89%; 2026-05-01: -64.71%; 2026-06-01: 50.00%; 2026-07-01: -22.22%; 2026-08-01: 14.29%; 2026-09-01: 71.43%-91.7354.28002024-10-01: -15.38 %2024-11-01: -27.27 %2024-12-01: 62.50 %2025-01-01: -23.08 %2025-02-01: -60.00 %2025-03-01: 150.00 %2025-04-01: -40.00 %2025-05-01: 16.67 %2025-06-01: -42.86 %2025-07-01: 150.00 %2025-08-01: 10.00 %2025-09-01: -27.27 %2025-10-01: -25.00 %2025-11-01: 216.67 %2025-12-01: -36.84 %2026-01-01: -91.67 %2026-02-01: 800.00 %2026-03-01: 0.00 %2026-04-01: 88.89 %2026-05-01: -64.71 %2026-06-01: 50.00 %2026-07-01: -22.22 %2026-08-01: 14.29 %2026-09-01: 71.43 %10/0109/01 UTC

Total 1,072 · Page 29/43 · Observed 9/27/2026, 3:30:50 AM

NameGroupReported valueDateNotesSource
BERT-LargeGoogle—2018-10-11Open weights (unrestricted)Source ↗
Transformer (Adaptive Input Embeddings) WT103Facebook AI Research—2018-09-28Open weights (unrestricted)Source ↗
LSTM+NeuralCacheKU Leuven,ESAT - PSI,Apple—2018-09-24UnreleasedSource ↗
AWD-LSTM-MoS + dynamic evaluation (WT2, 2018)Peking University,Microsoft Research Asia—2018-09-18UnreleasedSource ↗
Transformer + Simple Recurrent UnitASAPP,Cornell University,Google,Princeton University—2018-09-17UnreleasedSource ↗
NetSurfP-2.0Technical University of Denmark,University of Copenhagen,Universidad Nacional de San Martín,AIMST University—2018-09-10Hosted access (no API)Source ↗
(ensemble): AWD-LSTM-DOC (fin) × 5 (WT2)NTT Communication Science Laboratories,Tohoku University—2018-08-30Open weights (unrestricted)Source ↗
Big Transformer for Back-TranslationFacebook AI Research,Google Brain—2018-08-28Open weights (unrestricted)Source ↗
AWD-LSTM-MoS+PDR + dynamic evaluation (WT2)IBM—2018-08-14UnreleasedSource ↗
Big-Little Net (speech)IBM—2018-07-10Open weights (unrestricted)Source ↗
Big-Little NetIBM—2018-07-10Open weights (unrestricted)Source ↗
FTW (For The Win)DeepMind—2018-07-03UnreleasedSource ↗
MobileNetV2Google—2018-06-18—Source ↗
Relational Memory CoreDeepMind,University College London (UCL)—2018-06-05UnreleasedSource ↗
GPT-1OpenAI—2018-06-01Open weights (unrestricted)Source ↗
aLSTM(depth-2)+RecurrentPolicy (WT2)University of Manchester,Alan Turing Institute—2018-05-22UnreleasedSource ↗
Dropout-LSTM+Noise(Bernoulli) (WT2)Columbia University,New York University (NYU),Princeton University—2018-05-03UnreleasedSource ↗
ResNeXt-101 32x48dFacebook—2018-05-02Open weights (non-commercial)Source ↗
YOLOv3University of Washington—2018-04-08UnreleasedSource ↗
4 layer QRNN (h=2500)Salesforce Research—2018-03-22UnreleasedSource ↗
Chinese - English translationMicrosoft—2018-03-01—Source ↗
TCN (P-MNIST)Carnegie Mellon University (CMU),Intel Labs—2018-02-15—Source ↗
DeepLabV3+Google—2018-02-07—Source ↗
IMPALADeepMind—2018-02-05UnreleasedSource ↗
QRNNSalesforce Research—2018-02-01UnreleasedSource ↗

Data: Epoch AI · CC BY · View source · Automatically synced; not an exhaustive global sample