Organization
Carnegie Mellon University (CMU)
Tracked models
81
Frontier models
6
Largest training run
5.5 x 10^23 FLOP
81 tracked AI models list Carnegie Mellon University (CMU) as a developer. 6 of them are frontier models — among the largest training runs of their moment. The largest is LIMA, at 5.5 x 10^23 FLOP (May 2023). The most recent tracked release is from November 2025.
Models (80+)
LIMA5.5 x 10^23 FLOPLlemma 34B5.4 x 10^23 FLOPLlemma 7B1.2 x 10^23 FLOPStarCoder8.5 x 10^22 FLOPCRYSTALCODER5.6 x 10^22 FLOPAmber4.8 x 10^22 FLOPGigaGAN3.9 x 10^22 FLOPNoisy Student (L2)2.6 x 10^22 FLOPXLNet6.2 x 10^21 FLOPMamba-2.8B5.4 x 10^21 FLOPMamba 2, 2.7B4.9 x 10^21 FLOPAbGPT4.3 x 10^21 FLOPIncoder-6.7B3 x 10^21 FLOPSantaCoder2.1 x 10^21 FLOPHybrid-Phi-Mamba-1.5B1.2 x 10^21 FLOPPolyCoder1.1 x 10^21 FLOPJFT8.4 x 10^20 FLOPOcto-Base5.8 x 10^20 FLOPLibratus5.5 x 10^20 FLOPGemNet-OC5.4 x 10^20 FLOPTransformer-XL (257M)3.8 x 10^20 FLOPDEQ-Transformer (Post-LN) + Jacobian Regularisation2.9 x 10^19 FLOPNLM2.8 x 10^19 FLOPBiRNA-BERT1.8 x 10^19 FLOPHR-ResNet1017.1 x 10^18 FLOPHanabi 4 player4.3 x 10^18 FLOPAWD-LSTM-MoS + dynamic evaluation (WT2, 2017)3.4 x 10^18 FLOPTrellisNet2.8 x 10^18 FLOPDEQ-Transformer (Medium, Adaptive Embedding)8.2 x 10^17 FLOPDARTS3.2 x 10^17 FLOPTC-DNN-BLSTM-DNN1.9 x 10^17 FLOPPart-of-sentence tagging model1.5 x 10^17 FLOPNamed Entity Recognition model9.7 x 10^16 FLOPPeptideBERT4.9 x 10^16 FLOPENAS2 x 10^16 FLOPSystem 112.6 x 10^10 FLOPTranslation-invariant MLP1.8 x 10^10 FLOPALVINN1.1 x 10^10 FLOPMLP with back-propagation6.7 x 10^8 FLOPDistributed representation NN3.9 x 10^8 FLOPOlmo 3 32B InstructNovember 2025Olmo 3 7B ThinkNovember 2025Olmo 3.1 32B ThinkNovember 2025LoongRL 7BOctober 2025LoongRL 14BOctober 2025Octo-SmallMay 2024ManiGaussianMarch 2024AyaFebruary 2024VideoPoetDecember 2023Mamba-24M (SC09)December 2023NAEProOctober 2023GPT-MolBERTaSeptember 2023Robot ParkourSeptember 2023Innovative Drug-like Molecule Generation fromNovember 2022DLRM-12TApril 2021Transformer-XL + AutoDropout (WT2)January 2021SemExpJuly 2020MobileBERTApril 2020DEQ-TrellisNet (PTB)September 2019DEQ-TrellisNet (WT-103)September 2019SDEFebruary 2019Transformer-XL-ptbJanuary 2019TrellisNet-MoS (1.4x larger) PTBOctober 2018DARTS (second order) (PTB)June 2018LSTM (2018)March 2018TCN (148M)February 2018TCN (P-MNIST)February 2018TCN (13M)February 2018AWD-LSTM-MoS + dynamic evaluation (PTB, 2017)November 2017PhraseCondOctober 2017SESTMarch 2017Segmental RNNJune 2016Convolutional Pose MachinesJanuary 2016Listen, Attend and SpellAugust 2015Cross-Lingual POS TaggerJune 2011Boss (DARPA Urban Challenge)July 2008Sandstorm (DARPA Grand Challenge I)June 2004Probabilistic modeling for object recognitionJune 1998n-gram LMJuly 1997Time-delay neural networksMarch 1989
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.
Other organizations