Skip to content

NVIDIA RTX PRO 6000 Blackwell

NVIDIA · 96GB GDDR7 ECC · Can run 64 models

Buy Amazon
Manufacturer NVIDIA
VRAM 96 GB
Memory Type GDDR7 ECC
Architecture Blackwell
CUDA Cores 24,064
Tensor Cores 752
Bandwidth 1920 GB/s
TDP 600W
MSRP $6,800
Released Mar 18, 2025

AI Notes

The RTX PRO 6000 Blackwell is a top-tier workstation GPU with 96GB of ECC VRAM, making it one of the most capable single-GPU solutions for local AI. It can run 70B-parameter models at high quantizations and even fit some 100B+ models at Q4. The massive bandwidth and tensor core count deliver fast inference for production-grade AI deployments.

Compatible Models

Model Parameters Best Quant VRAM Used Fit Est. Speed
Qwen 3 0.6B 600M Q4_K_M 2.5 GB Runs ~768 tok/s
Gemma 3 1B 1B Q8_0 2 GB Runs ~960 tok/s
Llama 3.2 1B 1B Q8_0 3 GB Runs ~640 tok/s
DeepSeek R1 1.5B 1.5B Q8_0 3 GB Runs ~640 tok/s
Gemma 2 2B 2B Q8_0 4 GB Runs ~480 tok/s
Gemma 3n E2B 2B Q4_K_M 3.3 GB Runs ~582 tok/s
Llama 3.2 3B 3B Q8_0 5 GB Runs ~384 tok/s
Phi-3 Mini 3.8B 3.8B Q8_0 5.8 GB Runs ~331 tok/s
Phi-4 Mini 3.8B 3.8B Q4_K_M 4.5 GB Runs ~427 tok/s
Gemma 3 4B 4B Q4_K_M 5 GB Runs ~384 tok/s
Gemma 3n E4B 4B Q4_K_M 4.5 GB Runs ~427 tok/s
Qwen 3 4B 4B Q4_K_M 4.5 GB Runs ~427 tok/s
DeepSeek R1 7B 7B Q8_0 9 GB Runs ~213 tok/s
Falcon 3 7B 7B Q4_K_M 6.8 GB Runs ~282 tok/s
Mistral 7B 7B Q8_0 9 GB Runs ~213 tok/s
Qwen 2.5 7B 7B Q8_0 9 GB Runs ~213 tok/s
Qwen 2.5 Coder 7B 7B Q8_0 9 GB Runs ~213 tok/s
Qwen 2.5 VL 7B 7B Q4_K_M 7 GB Runs ~274 tok/s
Cogito 8B 8B Q4_K_M 7.5 GB Runs ~256 tok/s
DeepSeek R1 8B 8B Q4_K_M 7.5 GB Runs ~256 tok/s
Llama 3.1 8B 8B Q8_0 10 GB Runs ~192 tok/s
Nemotron 3 Nano 8B 8B Q4_K_M 7.5 GB Runs ~256 tok/s
Qwen 3 8B 8B Q4_K_M 7.5 GB Runs ~256 tok/s
Gemma 2 9B 9B Q8_0 11 GB Runs ~175 tok/s
Falcon 3 10B 10B Q4_K_M 8.5 GB Runs ~226 tok/s
Llama 3.2 Vision 11B 11B Q4_K_M 8.5 GB Runs ~226 tok/s
Gemma 3 12B 12B Q4_K_M 10.5 GB Runs ~183 tok/s
Mistral Nemo 12B 12B Q4_K_M 9.5 GB Runs ~202 tok/s
DeepSeek R1 14B 14B Q4_K_M 9.9 GB Runs ~194 tok/s
Phi-4 14B 14B Q4_K_M 9.9 GB Runs ~194 tok/s
Phi-4 Reasoning 14B 14B Q4_K_M 11 GB Runs ~175 tok/s
Qwen 2.5 14B 14B Q4_K_M 9.9 GB Runs ~194 tok/s
Qwen 2.5 Coder 14B 14B Q4_K_M 12 GB Runs ~160 tok/s
Qwen 3 14B 14B Q4_K_M 12 GB Runs ~160 tok/s
StarCoder2 15B 15B Q8_0 17 GB Runs ~113 tok/s
Codestral 22B 22B Q4_K_M 14.7 GB Runs ~131 tok/s
Devstral 24B 24B Q4_K_M 17 GB Runs ~113 tok/s
Magistral Small 24B 24B Q4_K_M 17 GB Runs ~113 tok/s
Mistral Small 3.1 24B 24B Q4_K_M 18 GB Runs ~107 tok/s
Gemma 2 27B 27B Q4_K_M 17.7 GB Runs ~108 tok/s
Gemma 3 27B 27B Q4_K_M 20 GB Runs ~96 tok/s
Qwen 3 30B-A3B (MoE) 30B Q4_K_M 22 GB Runs ~87 tok/s
Cogito 32B 32B Q4_K_M 21.5 GB Runs ~89 tok/s
DeepSeek R1 32B 32B Q4_K_M 20.7 GB Runs ~93 tok/s
Qwen 2.5 32B 32B Q4_K_M 20.7 GB Runs ~93 tok/s
Qwen 2.5 Coder 32B 32B Q4_K_M 23 GB Runs ~83 tok/s
Qwen 3 32B 32B Q4_K_M 23 GB Runs ~83 tok/s
QwQ 32B 32B Q4_K_M 21.5 GB Runs ~89 tok/s
Command R 35B 35B Q4_K_M 22.5 GB Runs ~85 tok/s
Mixtral 8x7B 47B Q4_K_M 29.7 GB Runs ~65 tok/s
Cogito 70B 70B Q4_K_M 43 GB Runs ~45 tok/s
DeepSeek R1 70B 70B Q4_K_M 43.5 GB Runs ~44 tok/s
Llama 3.1 70B 70B Q4_K_M 43.5 GB Runs ~44 tok/s
Llama 3.3 70B 70B Q4_K_M 43.5 GB Runs ~44 tok/s
Qwen 2.5 72B 72B Q4_K_M 44.7 GB Runs ~43 tok/s
Qwen 2.5 VL 72B 72B Q4_K_M 41 GB Runs ~47 tok/s
Llama 3.2 Vision 90B 90B Q4_K_M 50 GB Runs ~38 tok/s
Command R+ 104B 104B Q4_K_M 57 GB Runs ~34 tok/s
Llama 4 Scout (109B/17B active) 109B Q4_K_M 72 GB Runs ~27 tok/s
Command A 111B 111B Q4_K_M 61 GB Runs ~31 tok/s
Devstral 2 123B 123B Q4_K_M 67 GB Runs ~29 tok/s
Mistral Large 2 123B 123B Q4_K_M 67 GB Runs ~29 tok/s
Mixtral 8x22B 141B Q4_K_M 86 GB Runs (tight) ~22 tok/s
Qwen 3 235B-A22B 235B Q4_K_M 138 GB CPU Offload ~4 tok/s
5 model(s) are too large for this hardware.