Training › Model

Time to train versus accelerator count
Messaging Guidelines
Division
02004006008001,0001,20005001,000Accelerator countTime to train, minutesSlowerFaster

Results shown

146 of 146 results
Results shown
Plot on chart
Llama 3.1 8B3.947 min6.1-0133closedAvailable on-premise NVIDIA512128NVIDIA Rubin GPU4Olympus256NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B4.459 min6.0-0015closedAvailable Cloud NVIDIA1,024256NVIDIA Blackwell GPU (GB200)4Neoverse-V2512NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B4.636 min6.0-0101closedAvailable on-premise NVIDIA512128NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V2256NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B5.219 min5.1-0060closedAvailable on-premise NVIDIA512128NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V2256NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B5.432 min5.1-0072closedAvailable on-premise NVIDIA512128NVIDIA Blackwell GPU (GB200)4Neoverse-V2256NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B7.162 min6.1-0073closedAvailable on-premise CoreWeave7218NVIDIA Rubin GPU4Olympus36PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B7.380 min6.1-0123closedAvailable on-premise NVIDIA7218NVIDIA Rubin GPU4Olympus36NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B7.683 min6.1-0140closedAvailable on-premise Supermicro7218NVIDIA Rubin GPU4Olympus36NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B8.784 min6.1-0148closedAvailable on-premise University-of-Florida25632NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857064PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B10.024 min5.1-0087closedAvailable on-premise University-of-Florida25632NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857064PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B11.541 min6.1-0138closedAvailable on-premise Quanta_Cloud_Technology7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B11.569 min5.1-0085closedAvailable on-premise University-of-Florida12816NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857032PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B11.586 min6.0-0009closedAvailable Cloud Lambda7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236PyTorch NVIDIA Release 25.09Implementationv6.0v6.0
Llama 3.1 8B11.599 min6.1-0146closedAvailable on-premise University-of-Florida12816NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857032PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B11.753 min6.0-0096closedAvailable on-premise NVIDIA7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B11.821 min6.0-0077closedAvailable on-premise HPE7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA Release 26.04 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B11.844 min6.0-0063closedAvailable on-premise GigaComputing7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA NeMo Framework Release 25.09Implementationv6.0v6.0
Llama 3.1 8B11.860 min6.0-0110closedAvailable on-premise Supermicro7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NeMo Framework NVIDIA Release 26.04; NeMo Framework NVIDIA Release 26.03Implementationv6.0v6.0
Llama 3.1 8B11.872 min6.0-0026closedAvailable Cloud Nebius7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236Implementationv6.0v6.0
Llama 3.1 8B11.884 min6.1-0098closedAvailable on-premise HPE6416NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V232NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B12.068 min6.0-0030closedAvailable Cloud Oracle7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA NeMo Framework Release 25.09Implementationv6.0v6.0
Llama 3.1 8B12.159 min6.1-0137closedAvailable on-premise Quanta_Cloud_Technology6416NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V232NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B12.447 min6.0-0104closedAvailable on-premise Quanta_Cloud_Technology6416NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V232Implementationv6.0v6.0
Llama 3.1 8B12.686 min6.0-0045closedAvailable on-premise Dell6416NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V232PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B12.686 min6.1-0075closedAvailable on-premise Dell6416NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V232PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B12.976 min5.1-0057closedAvailable on-premise NVIDIA7218NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V236NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B13.701 min6.1-0070closedAvailable on-premise CoreWeave7218NVIDIA Blackwell GPU (GB200)4Neoverse-V236PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B14.074 min6.1-0003closedAvailable Cloud Crusoe12816AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F32flux1: PyTorch (ROCm 7.0.2) + TransformerEngine (fp8) MLPerf flux1 reference harness (torchtitan/NeMo); llama: AMD Primus (Megatron-LM pretrain / Megatron-Bridge post-train) from rocm/amd-mlperf:llama31_8b_training_6.0 and rocm/amd-mlperf:llama2_70b_training_6.0 (PyTorch 2.10 ROCm TransformerEngine ROCm AITER)Implementationv6.1v6.1
Llama 3.1 8B14.123 min6.1-0069closedAvailable on-premise Cisco648NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P16PyTorch NVIDIA Release 26.04 (Llama and GPT-OSS); 26.09 (DeepSeek and DLRMv4); 26.08 (Qwen) NVIDIA NeMo / Megatron TensorRT-LLM (Qwen generation) TorchRec (DLRMv4)Implementationv6.1v6.1
Llama 3.1 8B14.255 min5.1-0001closedAvailable Cloud Lambda7218NVIDIA Blackwell Ultra GPU (GB300)4Intel(R) Xeon(R) Platinum 857036PyTorch NVIDIA Release 25.04Implementationv5.1v5.1
Llama 3.1 8B14.416 min6.0-0025closedAvailable Cloud Nebius648NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P16NVIDIA NeMo Framework Release 25.09Implementationv6.0v6.0
Llama 3.1 8B15.020 min5.1-0063closedAvailable on-premise NVIDIA7218NVIDIA Blackwell GPU (GB200)4Neoverse-V236NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B16.127 min6.1-0150closedAvailable on-premise University-of-Florida648NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857016PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B16.536 min6.0-0002closedAvailable Cloud CoreWeave648NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 8562Y+16PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B18.100 min5.1-0090closedAvailable on-premise University-of-Florida648NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 857016PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B18.256 min5.1-0013closedAvailable Cloud Oracle648NVIDIA Blackwell GPU (B200-SXM-180GB)8INTEL(R) XEON(R) PLATINUM 8592+16MXNet NVIDIA Release 24.04 or PyTorch NVIDIA Release 24.04Implementationv5.1v5.1
Llama 3.1 8B18.305 min6.0-0034closedAvailable Cloud SCITIX648NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) 6767P16PyTorch NVIDIA Release 25.09Implementationv6.0v6.0
Llama 3.1 8B18.859 min5.1-0028closedAvailable on-premise Verda648NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 9575F16NeMo 2.1.0 Megatron-LM PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B20.200 min6.0-0060closedAvailable on-premise Dell328NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V216PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B20.200 min6.1-0090closedAvailable on-premise Dell328NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V216PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B21.433 min6.0-0076closedAvailable on-premise HPE328NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V216NVIDIA Release 26.04 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B21.975 min6.0-0042closedAvailable on-premise Cisco12816NVIDIA H100-SXM5-80GB8Intel(R) Xeon(R) Platinum 8592+32PyTorch NVIDIA NGC containerImplementationv6.0v6.0
Llama 3.1 8B24.748 min6.1-0149closedAvailable on-premise University-of-Florida324NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85708PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B27.828 min5.1-0007closedAvailable Cloud Nebius324NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85808NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B32.896 min5.1-0026closedAvailable on-premise Cisco12816NVIDIA H100-SXM5-80GB8Intel(R) Xeon(R) Platinum 8592+32PyTorch NVIDIA Release 23.09Implementationv5.1v5.1
Llama 3.1 8B33.391 min6.0-0075closedAvailable on-premise HPE164NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V28NVIDIA Release 26.04 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B33.808 min6.0-0044closedAvailable on-premise Cisco648NVIDIA H100-SXM5-80GB8Intel(R) Xeon(R) Platinum 8592+16PyTorch NVIDIA NGC containerImplementationv6.0v6.0
Llama 3.1 8B35.579 min6.0-0059closedAvailable on-premise Dell164NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V28PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B35.579 min6.1-0089closedAvailable on-premise Dell164NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V28PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B35.682 min5.1-0088closedAvailable on-premise University-of-Florida324NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85708PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B36.242 min6.1-0125closedAvailable on-premise NVIDIA82NVIDIA Rubin GPU4Olympus4NVIDIA NeMo 26.09 PlusImplementationv6.1v6.1
Llama 3.1 8B37.753 min5.1-0043closedAvailable on-premise HPE648NVIDIA H200-SXM5-141GB8Intel(R) Xeon(R) Platinum 8562Y+16NVIDIA Release 25.08 (NeMo PyTorch DGL HugeCTR)Implementationv5.1v5.1
Llama 3.1 8B39.014 min6.0-0072closedAvailable on-premise HPE328NVIDIA Blackwell GPU (GB200)4Neoverse-V216NVIDIA Release 25.09 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B40.964 min6.1-0100closedAvailable on-premise HPE162NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P4NVIDIA Release 26.04 (NeMo PyTorch DGL HugeCTR)Implementationv6.1v6.1
Llama 3.1 8B43.226 min6.1-0005closedAvailable Cloud Crusoe162AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F4gpt_oss_20b: AMD Primus (Megatron-LM pretrain) from rocm/amd-mlperf:llama31_8b_training_6.0 (PyTorch 2.10 ROCm TransformerEngine ROCm AITER); dlrmv4: PyTorch + TorchRec/FBGEMM-GPU (generative-recommenders HSTU) AMD MLPerf Training v6.1 dlrmv4 recipe; llama2_70b_lora: AMD Primus (Megatron-Bridge post-train) from rocm/amd-mlperf:llama2_70b_training_6.0 (PyTorch 2.10 ROCm TransformerEngine ROCm AITER); llama31_8b: AMD Primus (Megatron-LM pretrain) AMD MLPerf Training v6.1 llama31_8b recipeImplementationv6.1v6.1
Llama 3.1 8B43.317 min6.0-0058closedAvailable on-premise Dell162NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6747P4PyTorch NVIDIA Release 26.03Implementationv6.0v6.0
Llama 3.1 8B43.317 min6.1-0088closedAvailable on-premise Dell162NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P4PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B44.104 min6.1-0147closedAvailable on-premise University-of-Florida162NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85704PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B45.445 min5.1-0027closedAvailable on-premise Cisco648NVIDIA H100-SXM5-80GB8Intel(R) Xeon(R) Platinum 8592+16PyTorch NVIDIA Release 23.09Implementationv5.1v5.1
Llama 3.1 8B45.858 min6.0-0040closedAvailable on-premise Cisco162NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P4NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B49.047 min6.0-0071closedAvailable on-premise HPE164NVIDIA Blackwell GPU (GB200)4Neoverse-V28NVIDIA Release 25.09 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B51.827 min5.1-0006closedAvailable Cloud Nebius162NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85804NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B52.263 min6.0-0069closedAvailable on-premise HPE162NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 95354NVIDIA Framework Release 25.09 (NeMo PyTorch DGL HugeCTR)Implementationv6.0v6.0
Llama 3.1 8B52.402 min6.1-0094closedAvailable on-premise Firmus82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B55.050 min6.0-0109closedAvailable on-premise Supermicro162NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8AMD EPYC 9575F4NeMo Framework NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B56.074 min6.1-0152closedAvailable on-premise tinycorp162AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F4tinygradImplementationv6.1v6.1
Llama 3.1 8B58.629 min6.0-0043closedAvailable on-premise Cisco162AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F 64-Core Processor4PyTorch + ROCm (Primus / NeMo for LLM workloads)Implementationv6.0v6.0
Llama 3.1 8B59.561 min5.1-0086closedAvailable on-premise University-of-Florida162NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85704PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B60.811 min5.1-0038closedAvailable on-premise Dell162NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 99654NVIDIA NGC Pytorch 25.04 HugeCTR Nemo 25.09 DGL 25.03Implementationv5.1v5.1
Llama 3.1 8B62.053 min6.1-0054closedAvailable Cloud Oracle82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA PyTorch/NeMo Releases 26.09 and 26.10Implementationv6.1v6.1
Llama 3.1 8B62.643 min6.1-0139closedAvailable on-premise Quanta_Cloud_Technology82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA NeMo Framework Release 26.09Implementationv6.1v6.1
Llama 3.1 8B63.516 min6.0-0078closedAvailable on-premise HPE82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA Release 26.04 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B63.593 min6.0-0105closedAvailable on-premise Quanta_Cloud_Technology82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24Implementationv6.0v6.0
Llama 3.1 8B63.678 min6.0-0057closedAvailable on-premise Dell82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B63.678 min6.1-0087closedAvailable on-premise Dell82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B64.346 min6.0-0027closedAvailable Cloud Nebius82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24Implementationv6.0v6.0
Llama 3.1 8B65.258 min6.0-0031closedAvailable Cloud Oracle82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA NeMo Framework Release 25.09Implementationv6.0v6.0
Llama 3.1 8B66.042 min6.0-0098closedAvailable on-premise NVIDIA82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B67.373 min5.1-0058closedAvailable on-premise NVIDIA82NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V24NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B69.679 min6.1-0096closedAvailable on-premise GigaComputing81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6979P2PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B71.320 min6.1-0068closedAvailable on-premise ASUSTeK81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2NVIDIA Release pytorch 26.09Implementationv6.1v6.1
Llama 3.1 8B71.496 min6.1-0151closedAvailable on-premise University-of-Florida81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P2PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B71.600 min6.1-0084closedAvailable on-premise Dell81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8AMD EPYC 99652PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B72.012 min6.0-0023closedAvailable Cloud Nebius81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P2NVIDIA NeMo Framework Release 25.09Implementationv6.0v6.0
Llama 3.1 8B72.018 min6.1-0103closedAvailable on-premise Inventec81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6787P2NVIDIA Megatron-Bridge 26.08Implementationv6.1v6.1
Llama 3.1 8B72.942 min6.1-0062closedAvailable on-premise AMD81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2v26.6_primusImplementationv6.1v6.1
Llama 3.1 8B73.486 min6.0-0038closedAvailable on-premise ASUSTeK81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2NVIDIA Release pytorch 26.0426.03Implementationv6.0v6.0
Llama 3.1 8B73.691 min6.1-0008closedAvailable Cloud Crusoe81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P2NVIDIA NeMo Framework (Megatron-Bridge) built on NGC PyTorch 26.04 container (nvcr.io/nvidia/pytorch:26.04-py3)Implementationv6.1v6.1
Llama 3.1 8B74.081 min6.0-0066closedAvailable on-premise GigaComputing81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2PyTorchImplementationv6.0v6.0
Llama 3.1 8B74.255 min6.0-0053closedAvailable on-premise Dell81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6787P2PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B74.751 min6.0-0039closedAvailable on-premise Cisco81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B74.787 min6.0-0106closedAvailable on-premise Supermicro81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8AMD EPYC 9575F2NeMo Framework NVIDIA Release 26.04; Merlin HugeCTR NVIDIA Release 26.03Implementationv6.0v6.0
Llama 3.1 8B74.958 min6.0-0067closedAvailable on-premise HPE81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P2NVIDIA Release 26.04 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B75.132 min6.1-0086closedAvailable on-premise Dell81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 96552Pytorch 25.04Implementationv6.1v6.1
Llama 3.1 8B75.808 min6.0-0103closedAvailable on-premise Quanta_Cloud_Technology81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6740P2Implementationv6.0v6.0
Llama 3.1 8B75.841 min5.1-0008closedAvailable Cloud Nebius81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6776P2NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B76.165 min6.1-0141closedAvailable on-premise Supermicro81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2AMD Primus (PyTorch 2.12.0 ROCm 7.15.0)Implementationv6.1v6.1
Llama 3.1 8B76.407 min6.1-0102closedAvailable on-premise HPE81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 95552AMD Primus: Megatron-LM + Primus-Turbo (Llama 3.1 8B GPT-OSS 20B) Megatron-Bridge + Primus-Turbo (Llama 2 70B LoRA) all AMD MLPerf 6.1 partner code 253c561; AMD MLPerf v6.0 FLUX.1 implementation (FLUX.1); PyTorch + TorchRec + FBGEMM generative-recommenders HSTU (DLRMv4: AMD MLPerf 6.1 partner code 253c561)Implementationv6.1v6.1
Llama 3.1 8B76.768 min5.1-0050closedAvailable on-premise Lenovo81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6747P2PyTorch NVIDIA Release 23.09Implementationv5.1v5.1
Llama 3.1 8B77.328 min6.0-0079closedAvailable on-premise Inventec81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2NVIDIA Megatron-Bridge 26.04Implementationv6.0v6.0
Llama 3.1 8B77.393 min6.1-0101closedAvailable on-premise HPE81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8AMD EPYC 95552NVIDIA Framework Release 26.04 (NeMo PyTorch DGL HugeCTR)Implementationv6.1v6.1
Llama 3.1 8B78.339 min6.1-0004closedAvailable Cloud Crusoe81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2llama: AMD Primus (Megatron-Bridge post-train) from rocm/amd-mlperf:llama2_70b_training_6.0 (PyTorch 2.10 ROCm TransformerEngine ROCm AITER); dlrmv4: PyTorch + TorchRec/FBGEMM-GPU (generative-recommenders HSTU) AMD MLPerf Training v6.1 dlrmv4 recipe; llama31_8b: AMD Primus (Megatron-LM pretrain) AMD MLPerf Training v6.1 llama31_8b recipeImplementationv6.1v6.1
Llama 3.1 8B79.325 min5.1-0067closedAvailable on-premise NVIDIA82NVIDIA Blackwell GPU (GB200)4Neoverse-V24NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B79.692 min6.0-0074closedAvailable on-premise HPE82NVIDIA Blackwell GPU (GB200)4Neoverse-V24NVIDIA Release 25.09 (NeMo PyTorch)Implementationv6.0v6.0
Llama 3.1 8B80.907 min6.1-0083closedAvailable on-premise Dell81NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 96552PyTorch NVIDIA Release 26.09Implementationv6.1v6.1
Llama 3.1 8B81.894 min6.1-0145closedAvailable on-premise University-of-Florida81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85702PyTorch NVIDIA Release 26.09 (Flux.1: 26.08)Implementationv6.1v6.1
Llama 3.1 8B82.213 min6.0-0052closedAvailable on-premise Dell81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) 6767P2DGL NVIDIA Release 25.03Implementationv6.0v6.0
Llama 3.1 8B83.680 min6.0-0107closedAvailable on-premise Supermicro81NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 99652NeMo Framework NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B84.379 min5.1-0081closedAvailable on-premise Supermicro81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) 6787P2PyTorch NVIDIA Release 25.04; NeMo Framework NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B84.834 min5.1-0011closedAvailable Cloud Oracle81NVIDIA Blackwell GPU (B200-SXM-180GB)8INTEL(R) XEON(R) PLATINUM 8592+2MXNet NVIDIA Release 24.04 or PyTorch NVIDIA Release 24.04Implementationv5.1v5.1
Llama 3.1 8B85.027 min5.1-0030closedAvailable on-premise Dell81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 8562Y+2NVIDIA NGC Pytorch 25.04 HugeCTR Nemo 25.09 DGL 25.03Implementationv5.1v5.1
Llama 3.1 8B85.204 min5.1-0051closedAvailable on-premise Lenovo81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 8562Y+2PyTorch NVIDIA Release 23.09Implementationv5.1v5.1
Llama 3.1 8B85.263 min6.0-0008closedAvailable Cloud Lambda81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85702PyTorch NVIDIA Release 25.04Implementationv6.0v6.0
Llama 3.1 8B85.369 min5.1-0005closedAvailable Cloud Nebius81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85802NVIDIA NeMo Framework Release 25.09Implementationv5.1v5.1
Llama 3.1 8B85.371 min6.0-0068closedAvailable on-premise HPE81NVIDIA Blackwell GPU (B200-SXM-180GB)8AMD EPYC 95352NVIDIA Framework Release 25.09 (NeMo PyTorch DGL HugeCTR)Implementationv6.0v6.0
Llama 3.1 8B86.845 min6.0-0036closedAvailable on-premise AMD81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2AMD Primus Framework 0.2.0Implementationv6.0v6.0
Llama 3.1 8B90.037 min6.0-0108closedAvailable on-premise Supermicro81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2PyTorch 2.10.0+rocm7.2.2.git40d237bf ROCm 7.2.2Implementationv6.0v6.0
Llama 3.1 8B90.057 min6.0-0116closedAvailable on-premise vultr81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2PyTorch 2.10.0a0+git449b176 ROCm 7.2.0Implementationv6.0v6.0
Llama 3.1 8B91.145 min6.0-0033closedAvailable Cloud Oracle81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2ROCm 7.0Implementationv6.0v6.0
Llama 3.1 8B91.251 min6.0-0064closedAvailable on-premise GigaComputing81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2PyTorchImplementationv6.0v6.0
Llama 3.1 8B91.413 min6.0-0055closedAvailable on-premise Dell81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 96552Pytorch 25.04Implementationv6.0v6.0
Llama 3.1 8B91.939 min6.0-0070closedAvailable on-premise HPE81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 9575F2Implementationv6.0v6.0
Llama 3.1 8B92.075 min6.1-0067closedAvailable on-premise AMD81AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F2v26.6_primusImplementationv6.1v6.1
Llama 3.1 8B92.829 min6.0-0090closedAvailable on-premise MiTAC81AMD Instinct MI355X 288GB HBM3e8AMD EPYC 97552Pytorch 2.9.1 ROCm 7.2.2Implementationv6.0v6.0
Llama 3.1 8B93.687 min5.1-0084closedAvailable on-premise University-of-Florida81NVIDIA Blackwell GPU (B200-SXM-180GB)8Intel(R) Xeon(R) Platinum 85702PyTorch NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B94.817 min5.1-0077closedAvailable on-premise Supermicro81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8AMD EPYC 99652PyTorch NVIDIA Release 25.04; NeMo Framework NVIDIA Release 25.09Implementationv5.1v5.1
Llama 3.1 8B95.383 min5.1-0035closedAvailable on-premise Dell81NVIDIA Blackwell Ultra GPU (B300-SXM-270GB)8Intel(R) Xeon(R) 6767P2NVIDIA NGC Pytorch 25.04 HugeCTR Nemo 25.09 DGL 25.03Implementationv5.1v5.1
Llama 3.1 8B99.709 min5.1-0018closedAvailable on-premise AMD81AMD Instinct MI355X8AMD EPYC 9575F2PyTorch 2.8.0.dev20251001+rocm7.0.0.git64359f59 ROCm 7.0.0Implementationv5.1v5.1
Llama 3.1 8B101.278 min6.1-0153closedAvailable on-premise tinycorp81AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F2tinygradImplementationv6.1v6.1
Llama 3.1 8B103.559 min5.1-0076closedAvailable on-premise Supermicro81AMD Instinct MI355X8AMD EPYC 9575F2PyTorch ROCm 7.0.1Implementationv5.1v5.1
Llama 3.1 8B105.667 min5.1-0042closedAvailable on-premise GigaComputing81AMD Instinct MI355X8AMD EPYC 9575F2PyTorchImplementationv5.1v5.1
Llama 3.1 8B107.462 min5.1-0036closedAvailable on-premise Dell81AMD Instinct MI355X8AMD EPYC 96552Pytorch 25.04Implementationv5.1v5.1
Llama 3.1 8B108.965 min6.0-0092closedAvailable on-premise MiTAC81AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F2Flux1: PyTorch 2.10.0+rocm7.2.2.git40d237bf Llama3.1-8B:Pytorch 2.10.0a0+git449b176 ROCM 7.2.0Implementationv6.0v6.0
Llama 3.1 8B109.756 min6.0-0035closedAvailable on-premise AMD81AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F2AMD Primus Framework 0.2.0Implementationv6.0v6.0
Llama 3.1 8B122.199 min6.0-0041closedAvailable on-premise Cisco162NVIDIA H200-SXM5-141GB8AMD EPYC 9575F4NVIDIA NeMo Framework Release 26.04Implementationv6.0v6.0
Llama 3.1 8B122.929 min5.1-0017closedAvailable on-premise AMD81AMD Instinct MI350X8AMD EPYC 9575F2PyTorch 2.8.0.dev20251001+rocm7.0.0.git64359f59 ROCm 7.0.0Implementationv5.1v5.1
Llama 3.1 8B123.732 min6.0-0048closedAvailable on-premise Dell41NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V22PyTorch NVIDIA Release 26.04Implementationv6.0v6.0
Llama 3.1 8B123.732 min6.1-0078closedAvailable on-premise Dell41NVIDIA Blackwell Ultra GPU (GB300)4Neoverse-V22PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B127.625 min5.1-0025closedAvailable on-premise Cisco81AMD Instinct MI350X8AMD EPYC 9575F2PyTorch 2.8.0.dev20251001+rocm7.0.0.git64359f59 ROCm 7.0.0Implementationv5.1v5.1
Llama 3.1 8B187.479 min6.0-0115closedAvailable on-premise tinycorp81AMD Instinct MI350X 288GB HBM3e8AMD EPYC 9575F2tinygrad branch mlperf_training_v6.0Implementationv6.0v6.0
Llama 3.1 8B238.073 min6.0-0093closedAvailable on-premise MiTAC81AMD Instinct MI325X 256GB HBM3e8AMD EPYC 9575F2NeMo (flux1) / Primus + Megatron-LM (llama31_8b) on PyTorch ROCmImplementationv6.0v6.0
Llama 3.1 8B257.413 min6.0-0046closedAvailable on-premise Dell81NVIDIA H200-NVL-141GB8Intel(R) Xeon(R) 6767P2NVIDIA NGC Pytorch 26.04Implementationv6.0v6.0
Llama 3.1 8B259.950 min6.1-0077closedAvailable on-premise Dell81NVIDIA H200-NVL-141GB8Intel(R) Xeon(R) 6767P2NVIDIA NGC Pytorch 26.04Implementationv6.1v6.1
Llama 3.1 8B274.708 min6.1-0095closedAvailable on-premise Fujitsu81NVIDIA H200-NVL-141GB8Intel(R) Xeon(R) Platinum 8592+2PyTorch NVIDIA Release 26.04Implementationv6.1v6.1
Llama 3.1 8B375.636 min6.0-0114closedAvailable on-premise TTA81NVIDIA RTX PRO 6000 Blackwell Server Edition8AMD EPYC 96542NeMo 2.6.0rc0 + Megatron-LM core_r0.11.0Implementationv6.0v6.0
Llama 3.1 8B1,187.380 min6.1-0144closedAvailable on-premise TTA41H100 NVL Tensor Core GPU (PCIe)4Intel(R) Xeon(R) Platinum 85582PyTorch NVIDIA Release 25.09 (NeMo)Implementationv6.1v6.1