Can NVIDIA GeForce RTX 5090 run Qwen 2.5 72B?

72B parameter LLM model on 32GB GDDR7

Barely — requires CPU/RAM offloading
~1-3 tok/s (offload)
SpeedVery slow — expect 1-3 tokens/sec
QualityQuality is fine but the speed makes it impractical for interactive use

VRAM Requirements

Qwen 2.5 72B is a 72B parameter model. At full precision (FP16), it requires 144GB of VRAM. Your NVIDIA GeForce RTX 5090 only has 32GB — not enough even at maximum compression.

FP16 (Full Precision)144GB (need 112GB more)

Maximum quality, no quantization

Q8 (8-bit)72GB (need 40GB more)

Near-lossless, ~50% size reduction

Q4 (4-bit)42GB (need 10GB more)

Good quality, ~75% size reduction

Your GPU VRAM: 32GB GDDR7 at 1792 GB/s bandwidth
Recommended system RAM: 64GB DDR5 (2x GPU VRAM minimum for model overflow)

How to Set It Up

Step 1: Install Ollama

curl -fsSL https://ollama.com/install.sh | sh

Ollama is the easiest way to run local LLMs. Works on Linux, macOS, and Windows.

Step 2: Download and run Qwen 2.5 72B

ollama run qwen2.5:72b:q4_K_M

This downloads the model (~72GB). First run takes a few minutes.

Step 3: Verify GPU is being used

nvidia-smi

Check that VRAM usage increases when the model loads. You should see ~42GB used.

NVIDIA GeForce RTX 5090 Specs

VRAM32GB GDDR7
Memory Bandwidth1792 GB/s
TDP575W
CUDA Cores21,760
Street Price~$2800
AI Rating10/10

About Qwen 2.5 72B

Top open LLM for reasoning. Similar requirements to Llama 70B.

Category: LLM · Parameters: 72B · CUDA required: No (runs via llama.cpp/GGUF)