Country
China
Tracked models
916
Frontier models
7
Largest training run
2.5 x 10^25 FLOP
Data-center compute
278.5K H100e
916 tracked AI models have at least one contributing organization based in China. 7 of them are frontier models — among the largest training runs of their moment. The largest is Doubao-pro, at 2.5 x 10^25 FLOP (October 2024). The most recent tracked release is from July 2026.
3 tracked AI data centers sit on its territory, holding about 278.5K H100e and drawing 632 MW. It hosts 30 tracked GPU clusters.
Models (80+)
Doubao-pro2.5 x 10^25 FLOPKimi K32 x 10^25 FLOPQwen3-Max1.5 x 10^25 FLOPPangu Ultra1.1 x 10^25 FLOPEmu3.59.9 x 10^24 FLOPDeepSeek-V4-Pro9.7 x 10^24 FLOPQwen2.5-VL-72B9.6 x 10^24 FLOPQwen2.5 Instruct (72B)7.9 x 10^24 FLOPQwen2.5-72B7.8 x 10^24 FLOPQwen3-32B7.1 x 10^24 FLOPTelechat2-115B6.9 x 10^24 FLOPGLM-56.8 x 10^24 FLOPMiMo-V2.5-Pro6.8 x 10^24 FLOPLing-1T6 x 10^24 FLOPKimi K2.55.8 x 10^24 FLOPHunyuan-TurboS5.4 x 10^24 FLOPQwen3-235B-A22B-Thinking (Jul 2025)4.8 x 10^24 FLOPQwen3-235B-A22B (Jul 2025)4.8 x 10^24 FLOPQwen3-235B-A22B4.8 x 10^24 FLOPGLM-4.74.4 x 10^24 FLOPGLM-4.64.4 x 10^24 FLOPGLM-4.54.4 x 10^24 FLOPMiniMax-M1-80k4.3 x 10^24 FLOPMiMo-V2.54.3 x 10^24 FLOPDeepSeek-V3.24.2 x 10^24 FLOPKimi K2 Thinking4.2 x 10^24 FLOPMiniMax-M1-40k4.2 x 10^24 FLOPDeepSeek-V3.2-Exp4.2 x 10^24 FLOPStep-Video-T2V4.1 x 10^24 FLOPDeepSeek-R1 (May 2025)4 x 10^24 FLOPMegaScale (Production)3.9 x 10^24 FLOPLongCat-Flash3.7 x 10^24 FLOPDeepSeek-V3.1-Terminus3.6 x 10^24 FLOPDeepSeek-V3.13.6 x 10^24 FLOPQwQ-32B3.5 x 10^24 FLOPQwen2.5-32B3.5 x 10^24 FLOPDeepSeek-R13.5 x 10^24 FLOPHunyuan-Large3.5 x 10^24 FLOPDeepSeek-V3 (Mar 2025)3.3 x 10^24 FLOPDeepSeek-V33.3 x 10^24 FLOPQwen3-14B3.2 x 10^24 FLOPMiniMax-Text-013.1 x 10^24 FLOPPangu Ultra MoE3.1 x 10^24 FLOPLLaVA-OV-72B3 x 10^24 FLOPQwen2-72B3 x 10^24 FLOPKimi K23 x 10^24 FLOPGLM-Z1-Rumination-32B-04142.9 x 10^24 FLOPGLM-4-32B-04142.9 x 10^24 FLOPERNIE-4.5-300B-A47B2.8 x 10^24 FLOPSeed-OSS-36B-Base2.6 x 10^24 FLOPStep-Omni2.5 x 10^24 FLOPDeepSeek-V4-Flash2.5 x 10^24 FLOPMiMo-V2-Flash2.4 x 10^24 FLOPMiniMax-VL-012.1 x 10^24 FLOPStep-Audio-Chat 130B1.9 x 10^24 FLOPGLM-4.5V1.8 x 10^24 FLOPYi-Large1.8 x 10^24 FLOPDeepSeek-V2.51.8 x 10^24 FLOPQwen3-8B1.8 x 10^24 FLOPGLM-4.5-Air1.7 x 10^24 FLOPQwen2.5-14B1.6 x 10^24 FLOPQwen3-Coder-480B-A35B1.6 x 10^24 FLOPLing-Plus ("Bailing")1.6 x 10^24 FLOPWu Dao 2.01.5 x 10^24 FLOPYi-Lightning1.5 x 10^24 FLOPSeed1.5-VL1.4 x 10^24 FLOPQwen1.5-72B1.3 x 10^24 FLOPQwen-72B1.3 x 10^24 FLOPPangu Pro MoE1.3 x 10^24 FLOPDeepSeek-Coder-V2 236B1.3 x 10^24 FLOPXVERSE-65B-21.2 x 10^24 FLOPdots.llm11.2 x 10^24 FLOPHunyuan1.2 x 10^24 FLOPLuca 2.01.2 x 10^24 FLOPMiMo-VL-7B-SFT1.2 x 10^24 FLOPQwen2.5-Coder (32B)1.1 x 10^24 FLOPMiMo-7B-Base1.1 x 10^24 FLOPBlueLM 175B1.1 x 10^24 FLOPERNIE 3.0 Titan10^24 FLOPDeepSeek-V2 (MoE-236B)10^24 FLOP
Data centers (3)
GPU clusters (30)
Anonymized Chinese System10K H100eAnonymized Chinese System10K H100eAnonymized Chinese System8K H100eAnonymized Chinese System8K H100eAnonymized Chinese System6K H100eAnonymized Chinese System5K H100eAnonymized Chinese System5K H100eAnonymized Chinese System4K H100eAnonymized Chinese System4K H100eAnonymized Chinese System4K H100eAnonymized Chinese System3K H100eAnonymized Chinese System3K H100eAnonymized Chinese System3K H100eAnonymized Chinese System3K H100eAnonymized Chinese System3K H100eAnonymized Chinese System2K H100eAnonymized Chinese System2K H100eAnonymized Chinese System2K H100eAnonymized Chinese System2K H100eAnonymized Chinese System2K H100eAnonymized Chinese System1K H100eAnonymized Chinese System1K H100eAnonymized Chinese System1K H100eAnonymized Chinese System1K H100eAnonymized Chinese System1K H100eAnonymized Chinese System1K H100eAnonymized Chinese System900 H100eAnonymized Chinese System800 H100eAnonymized Chinese System600 H100eAnonymized Chinese System500 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.
Other countries