Hardware / NVIDIA /H200 141GB SXM
NVIDIA · GPU

H200 141GB SXM

Runs 146 models fully in memory, 14 more with CPU offload. 2 of those sit close enough to 141 GB that our size estimate is what decides them.

Memory
141 GB
Bandwidth
4800 GB/s
Runs
146
01

Specifications

Memory
141 GB
Bandwidth
4800 GB/s
FP16 compute
989 TFLOPS
TDP
700 W
MSRP
$35,000
Street price
Released
Apr 1, 2024
Architecture
Hopper
02

What it runs

Nex-N2-Mini35.1BQ5_K_M262K131 tok/sestFits in memoryGranite 4.1 8B8.8BQ8_0131K411 tok/sestFits in memoryQwen3.6 35B A3B36BQ5_K_M262K1530 tok/sestFits in memoryGemma 4 26B A4B26.5BQ8_0262K768 tok/sestFits in memoryGemma 4 31B32.7BQ5_K_M66K165 tok/sestFits in memoryQwen3.5-9B9.7BQ5_K_M262K557 tok/sestFits in memoryQwen3.5-27B27.8BQ5_K_M262K194 tok/sestFits in memoryQwen3.5-122B-A10B125BQ5_K_M262K459 tok/sestFits in memoryGLM 4.7 Flash31.2BQ4_K_M131K173 tok/sestFits in memoryNemotron 3 Nano 30B A3B31.6BQ5_K_M262K1530 tok/sestFits in memoryMinistral 3 14B 251213.9BQ5_K_M262K388 tok/sestFits in memoryMinistral 3 8B 25128.9BQ8_0262K406 tok/sestFits in memoryMinistral 3 3B 25123.8BQ4_K_M131K1667 tok/sestFits in memoryOlmo 3 32B Think32.2BQ4_K_M66K197 tok/sestFits in memorygpt-oss-safeguard-20b21.5BQ4_K_M131K250 tok/sestFits in memoryQwen3 VL 32B Instruct33.4BQ5_K_M131K162 tok/sestFits in memoryGranite 4.0 Micro3.2BQ5_K_M66K1434 tok/sestFits in memoryQwen3 VL 8B Thinking8.8BQ8_0131K411 tok/sestFits in memoryQwen3 VL 30B A3B Thinking31.1BQ5_K_M262K1530 tok/sestFits in memoryCydonia 24B V4.123.6BQ5_K_M131K229 tok/sestFits in memoryQwen3 Next 80B A3B Thinking81.3BQ4_K_M262K1795 tok/sestFits in memoryUI-TARS 7B8.3BQ5_K_M66K651 tok/sestFits in memoryUncensored24BQ8_066K151 tok/sestFits in memoryHunyuan A13B Instruct80.4BQ8_0131K236 tok/sestFits in memoryMistral Small 3.2 24B24BQ4_K_M131K264 tok/sestFits in memoryLlama Guard 4 12B12BQ4_K_M262K528 tok/sestFits in memoryQwen3 14B14.8BQ4_K_M131K428 tok/sestFits in memoryGemma 3 12B12.2BQ5_K_M131K443 tok/sestFits in memoryGemma 3 27B27.4BQ4_K_M262K231 tok/sestFits in memoryRocinante 12B12.2BQ5_K_M66K443 tok/sestFits in memoryLlama 3.2 3B Instruct3.2BQ8_0131K1129 tok/sestFits in memoryMistral Nemo12.2BQ4_K_M131K519 tok/sestFits in memoryMixtral 8x22B Instruct141BQ4_K_M66K138 tok/sestFits in memoryLaguna XS 2.133.4BQ4_K_M262K161 tok/sestFits in memoryQwen3.6 27B27.8BQ4_K_M262K228 tok/sestFits in memoryQwen3.5-35B-A3B36BQ4_K_M262K1795 tok/sestFits in memoryQwen3 VL 8B Instruct8.8BQ4_K_M262K720 tok/sestFits in memoryQwen3 VL 30B A3B Instruct31.1BQ8_0262K1024 tok/sestFits in memoryQwen3 Next 80B A3B Instruct81.3BQ5_K_M262K1530 tok/sestFits in memoryQwen3 30B A3B Thinking 250730.5BQ5_K_M66K1530 tok/sestFits in memoryQwen3 Coder 30B A3B Instruct30.5BQ4_K_M262K1795 tok/sestFits in memoryQwen3 30B A3B Instruct 250730.5BQ8_0262K1024 tok/sestFits in memoryQwen3 8B8.2BQ8_0131K441 tok/sestFits in memoryQwen3 32B32.8BQ5_K_M131K165 tok/sestFits in memoryMistral Small 3.1 24B24BQ8_066K151 tok/sestFits in memoryGemma 3 4B4.3BQ8_0131K841 tok/sestFits in memoryReka Flash 320.9BQ8_066K173 tok/sestFits in memoryLlama 3.1 8B Instruct8BQ8_0131K452 tok/sestFits in memoryNiagara 19m Batch.en20MQ8_0~262K180706 tok/sestFits in memoryNiagara 38m Batch.en38MQ5_K_M~262K142078 tok/sestFits in memoryARK ASR 0.6B1.3BQ5_K_M~262K4153 tok/sestFits in memoryARK ASR 3B4.1BQ8_0~262K882 tok/sestFits in memoryAudio8 ASR 0.1B0.3BQ5_K_M~262K17997 tok/sestFits in memoryHiggs Audio v3 8b STT v28.9BQ4_K_M~262K712 tok/sestFits in memoryHiggs Audio v3 STT2.7BQ5_K_M~262K2000 tok/sestFits in memoryCohere Transcribe 03 20262.1BQ5_K_M~262K2571 tok/sestFits in memoryDistil Large v3.50.8BQ8_0~262K4518 tok/sestFits in memoryLite Whisper Large v3 Acc1.4BQ8_0~262K2582 tok/sestFits in memoryOwsm CTC v3.1 1B1.1BQ4_K_M~262K5758 tok/sestFits in memoryOwsm CTC v3.2 ft 1B1BQ8_0~262K3614 tok/sestFits in memory

A speed marked est is calculated from this card's memory bandwidth and the size of the weights it has to read per word. It is not a measurement.

Something wrong on this page? Tell us