United States

Tracked models

1,909

Frontier models

95

Largest training run

5 x 10^26 FLOP

Data-center compute

9.8M H100e

1,909 tracked AI models have at least one contributing organization based in United States. 95 of them are frontier models — among the largest training runs of their moment. The largest is Grok 4, at 5 x 10^26 FLOP (July 2025). The most recent tracked release is from July 2026.

30 tracked AI data centers sit on its territory, holding about 9.8M H100e and drawing 8.82 GW. It hosts 30 tracked GPU clusters.

Models (80+)

Grok 45 x 10^26 FLOP GPT-4.53.8 x 10^26 FLOP Grok 33.5 x 10^26 FLOP GPT-56.6 x 10^25 FLOP Llama 4 Behemoth (preview)5.2 x 10^25 FLOP Gemini 1.0 Ultra5 x 10^25 FLOP Llama Nemotron Ultra 253B3.9 x 10^25 FLOP Composer 2.53.9 x 10^25 FLOP Llama 3.1-405B3.8 x 10^25 FLOP Claude 3.7 Sonnet3.4 x 10^25 FLOP Grok-23 x 10^25 FLOP Claude 3.5 Sonnet2.7 x 10^25 FLOP Composer 22.3 x 10^25 FLOP GPT-4 (Jun 2023)2.1 x 10^25 FLOP GPT-4 (Mar 2023)2.1 x 10^25 FLOP Nemotron-4 340B1.8 x 10^25 FLOP Inflection-210^25 FLOP Grok-1.59.3 x 10^24 FLOP Reka Core8.4 x 10^24 FLOP Inflection-2.58 x 10^24 FLOP Llama 3.1-70B7.9 x 10^24 FLOP Llama-3.1-Nemotron-70B-Instruct7.9 x 10^24 FLOP Llama 3-70B7.9 x 10^24 FLOP AMIE (Articulate Medical Intelligence Explorer)7.3 x 10^24 FLOP PaLM 27.3 x 10^24 FLOP MAI-Thinking-17 x 10^24 FLOP Llama 3.3 70B6.9 x 10^24 FLOP Nemotron-H 56B6.7 x 10^24 FLOP Nemotron 3 Ultra6.6 x 10^24 FLOP Amazon Nova Pro6 x 10^24 FLOP gpt-oss-120b4.9 x 10^24 FLOP Amazon Titan4.8 x 10^24 FLOP AFM-server4.3 x 10^24 FLOP Llama 4 Scout4.1 x 10^24 FLOP Claude 23.9 x 10^24 FLOP Cosmos-Transfer1-7B3.6 x 10^24 FLOP NVLM-X 72B3 x 10^24 FLOP NVLM-H 72B3 x 10^24 FLOP NVLM-D 72B3 x 10^24 FLOP Grok-12.9 x 10^24 FLOP Cosmos-1.0- Diffusion-14B Video2World2.8 x 10^24 FLOP Cosmos-Predict1-14b-Video2World2.8 x 10^24 FLOP Minerva (540B)2.7 x 10^24 FLOP DBRX2.6 x 10^24 FLOP GPT-3.5 (davinci-002)2.6 x 10^24 FLOP Flan-PaLM 540B2.5 x 10^24 FLOP U-PaLM (540B)2.5 x 10^24 FLOP PaLM (540B)2.5 x 10^24 FLOP Gemma 3 27B2.3 x 10^24 FLOP Evo 2 40B2.3 x 10^24 FLOP Llama 4 Maverick2.2 x 10^24 FLOP TxGemma 27B2.1 x 10^24 FLOP Gemma 2 27B2.1 x 10^24 FLOP FLAN 137B2 x 10^24 FLOP Inkling1.8 x 10^24 FLOP Movie Gen Video1.6 x 10^24 FLOP Chameleon-34B1.6 x 10^24 FLOP NVIDIA-Nemotron-Nano-9B-v21.5 x 10^24 FLOP NVIDIA-Nemotron-Nano-12B-v21.5 x 10^24 FLOP Cosmos-Predict1-7b-Video2World1.4 x 10^24 FLOP Arcee Trinity Large1.3 x 10^24 FLOP OLMo 2 32B1.3 x 10^24 FLOP Code Llama-70B1.3 x 10^24 FLOP Foundation-sec-8b1.2 x 10^24 FLOP Llama 3.1-8B1.2 x 10^24 FLOP Granite-4.0-H-Small1.2 x 10^24 FLOP Olmo 31.1 x 10^24 FLOP ESM3 (98B)1.1 x 10^24 FLOP Inflection-110^24 FLOP Phi-4-Reasoning9.3 x 10^23 FLOP Phi-49.3 x 10^23 FLOP Gemma 3 12B8.6 x 10^23 FLOP Megatron-Turing NLG 530B8.6 x 10^23 FLOP SambaLingo-Thai-Chat-70B8.1 x 10^23 FLOP Llama 2-70B8.1 x 10^23 FLOP Nemotron-4 15B7.5 x 10^23 FLOP Nemotron-H 8B7.2 x 10^23 FLOP Llama 3-8B7.2 x 10^23 FLOP Reka Flash6.3 x 10^23 FLOP Marin 8B6.1 x 10^23 FLOP

Data centers (30)

Colossus 21.1M H100e Microsoft Fairwater Atlanta768.8K H100e Meta Prometheus763K H100e Anthropic-Amazon New Carlisle685.9K H100e OpenAI Stargate Abilene509.3K H100e Microsoft Fairwater Wisconsin445.7K H100e Google Columbus408.8K H100e Google New Albany351.7K H100e Google Council Bluffs (East)335K H100e Google Lincoln287.5K H100e Google Bristow283.7K H100e Colossus 1275.8K H100e Google Omaha267.8K H100e CoreWeave Denton TX252.7K H100e QTS Richmond 1242.5K H100e Meta Rosemount215.8K H100e Amazon Madison Mega Site214.3K H100e Microsoft Goodyear205.2K H100e Meta Jeffersonville203.1K H100e Google The Dalles201.6K H100e Google Midlothian191K H100e QTS Richmond 2182.9K H100e QTS Richmond 3174.8K H100e Meta Montgomery174.5K H100e Meta Kuna173.3K H100e Meta Temple173.2K H100e Meta Cheyenne172.8K H100e Amazon Ridgeland171.3K H100e Meta Aiken161.7K H100e Coreweave Helios159.6K H100e

GPU clusters (30)

xAI Colossus Memphis Phase 3275.8K H100e xAI Colossus Memphis Phase 2200K H100e xAI Colossus Memphis Phase 1100K H100e OpenAI/Microsoft Goodyear Arizona100K H100e Meta 100k100K H100e Oracle OCI Supercluster H200s65.5K H100e Tesla Cortex Phase 150K H100e Lawrence Livermore NL El Capitan Phase 244.1K H100e CoreWeave H200s42K H100e Meta GenAI 2024b24.6K H100e Meta GenAI 2024a24.6K H100e Oracle OCI MI300x21.6K H100e Oracle OCI Supercluster H100s16.4K H100e Microsoft Azure Eagle14.4K H100e NVIDIA CoreWeave Eos-DFW Phase 110.8K H100e Oracle OCI Supercluster A100s10.3K H100e Imbue 10k Cluster10K H100e Tesla 10k H100 Cluster10K H100e Magic G4 Google Cloud Rental8K H100e Microsoft GPT-4 cluster7.9K H100e Oak Ridge NL Frontier7.3K H100e Meta Research SuperCluster (RSC-1) Phase 25K H100e NVIDIA Eos Phase 24.6K H100e Lawrence Livermore NL Tuolumne4.6K H100e Google Oklahoma TPU v4 Pods4.6K H100e Together AI H100 Cluster4.4K H100e Amazon Titan training cluster4.3K H100e ExxonMobil Discovery 64K H100e Microsoft Azure ND H100 v5 VM4K H100e CoreWeave LiquidLab4K H100e

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.