RTX Pro 5000 Blackwell

29 records across 7 models (quants: MXFP4, Q4_0, Q4_K).

hardwaremodelquantbackendtok/spp tok/sWttft sctx/tgdateclassflagssource
RTX Pro 5000 Blackwellgpt-oss 20B 20BMXFP4llama.cpp213.410339.6––4096–sourced–source
RTX Pro 5000 Blackwellgpt-oss 20B 20BMXFP4llama.cpp196.28747.7––16384–sourced–source
RTX Pro 5000 Blackwellgpt-oss 20B 20BMXFP4llama.cpp182.87257.7––32768–sourced–source
RTX Pro 5000 Blackwellgpt-oss 20B 20BMXFP4llama.cpp161.45273.9––65536–sourced–source
RTX Pro 5000 Blackwellgpt-oss 20B 20BMXFP4llama.cpp130.93163.3––131072–sourced–source
RTX Pro 5000 BlackwellLlama 3.3 70B 70BQ4_Kllama.cpp24.01022.9––4096–sourced–source
RTX Pro 5000 BlackwellLlama 3.3 70B 70BQ4_Kllama.cpp22.4826.3––16384–sourced–source
RTX Pro 5000 Blackwellllama 7B Q4_0llama.cpp (CUDA)226.3111046.55––tg1282026-02-28community–source
RTX Pro 5000 Blackwellllama 7B Q4_0llama.cpp (CUDA)213.549952.91––tg1282026-02-28community–source
RTX Pro 5000 BlackwellQwen3 14B 14BQ4_Kllama.cpp98.94515.7––4096–sourced–source
RTX Pro 5000 BlackwellQwen3 14B 14BQ4_Kllama.cpp84.33064.2––16384–sourced–source
RTX Pro 5000 BlackwellQwen3 14B 14BQ4_Kllama.cpp70.82069.8––32768–sourced–source
RTX Pro 5000 BlackwellQwen3 14B 14BQ4_Kllama.cpp53.81125.1––65536–sourced–source
RTX Pro 5000 BlackwellQwen3 14B 14BQ4_Kllama.cpp36.5434.2––131072–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp188.85266.1––4096–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp148.54043.8––16384–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp123.23032.7––32768–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp91.91708.9––65536–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp61.4702.7––131072–sourced–source
RTX Pro 5000 BlackwellQwen3 30B A3B 30BQ4_Kllama.cpp37.0364.1––262144–sourced–source
RTX Pro 5000 BlackwellQwen3 32B 32BQ4_Kllama.cpp46.62030.3––4096–sourced–source
RTX Pro 5000 BlackwellQwen3 32B 32BQ4_Kllama.cpp42.01497.7––16384–sourced–source
RTX Pro 5000 BlackwellQwen3 32B 32BQ4_Kllama.cpp36.41079.8––32768–sourced–source
RTX Pro 5000 BlackwellQwen3 32B 32BQ4_Kllama.cpp28.9629.6––65536–sourced–source
RTX Pro 5000 BlackwellQwen3 8B 8BQ4_Kllama.cpp159.88175.3––4096–sourced–source
RTX Pro 5000 BlackwellQwen3 8B 8BQ4_Kllama.cpp126.35846.4––16384–sourced–source
RTX Pro 5000 BlackwellQwen3 8B 8BQ4_Kllama.cpp98.34006.2––32768–sourced–source
RTX Pro 5000 BlackwellQwen3 8B 8BQ4_Kllama.cpp64.12120.7––65536–sourced–source
RTX Pro 5000 BlackwellQwen3 8B 8BQ4_Kllama.cpp38.1899.1––131072–sourced–source

All rows cite their source. Provenance: community 40018 sourced 1085

← all hardware