RTX PRO 5000 72 GB Blackwell

The RTX PRO 5000 72 GB Blackwell is an NVIDIA GPU built on the Blackwell 2.0 architecture, released 2025-10-21. For running AI locally, the numbers that matter are its 72 GB of GDDR7 and 1340 GB/s of memory bandwidth. VRAM decides which models fit at all; bandwidth sets how fast they generate text.

What you can run on 72 GB

At Q4_K_M quantization (the usual local default), 72 GB holds models up to roughly 114B parameters, leaving headroom for context. On this card you can run, among others:

• Qwen3-Next-80B-A3B-Instruct - 81.3B parameters
• Qwen3-Coder-Next - 79.7B parameters
• Qwen2.5 72B - 72B parameters
• Llama-3.1-70B-Instruct - 70.6B parameters
• Llama 3.3 70B - 70B parameters

Larger models need a higher-VRAM card, a second GPU, or CPU offload (which is much slower). Check any specific model with the VRAM calculator, or see the full picture on what can I run.

Local LLM speed (LLaMA 3, llama.cpp)

Single-stream token-generation throughput - estimated from memory bandwidth:

Model (quant)	Speed on RTX PRO 5000 72 GB Blackwell
Llama 3 8B (Q4_K_M)	156.8 tok/s
Llama 3 8B (F16)	71.2 tok/s
Llama 3 70B (Q4_K_M)	18.4 tok/s

Because decode is memory-bandwidth bound, the 1340 GB/s figure is the best single predictor of chat speed on this card. Estimates are calibrated against measured RTX-40-series cards and are typically within ~15%.

Memory and power

• VRAM: 72 GB GDDR7 (384-bit bus)
• Bandwidth: 1340 GB/s
• TDP: 300 W - a 700 W+ power supply is recommended
• Process: 5 nm
• Interface: PCIe 5.0 x16

Quantization and context

Quantization trades a little quality for a lot of VRAM. On 72 GB you can fit roughly a 114B model at Q4_K_M, about a 61B model at the higher-quality Q8, or a smaller model at full FP16. Longer context windows also consume VRAM (the KV cache grows with context length), so leave a few GB of headroom if you plan to use large prompts or many concurrent requests. For most chat and coding use, Q4_K_M on this card is the sweet spot between speed, quality, and the 72 GB budget.

How it compares

Similar cards for local AI, by VRAM and 8B-Q4 speed:

GPU	VRAM	Bandwidth	Llama 3 8B Q4
RTX PRO 5000 72 GB Blackwell	72 GB	1340 GB/s	156.8 tok/s
NVIDIA RTX A6000	48 GB	768 GB/s	89.9 tok/s
NVIDIA L40S	48 GB	864 GB/s	101.1 tok/s
NVIDIA RTX 6000 Ada Generation	48 GB	960 GB/s	112.3 tok/s

Bottom line

The RTX PRO 5000 72 GB Blackwell is best for llm-inference, image-gen, comfyui. With 24 GB+ it comfortably handles most open models, including 30B-class at Q4. If you need more, compare with NVIDIA RTX A6000 and NVIDIA L40S.

Sources

• Specifications: RightNow GPU Database (TechPowerUp data)
• Benchmarks: GPU-Benchmarks-on-LLM-Inference (basis for the bandwidth estimate)

Specs and benchmarks last checked 2026-06-08. Verify current pricing before buying.