Can the DGX Spark (GB10) run Qwen2.5 72B?

Yes. Qwen2.5 72B fits the DGX Spark (GB10)'s 128 GB at Q4_K_M (about 41.8 GB needed), running at an estimated 3.9 tok/s. There's room for up to Q8_0 if you want higher quality.

VRAM
128 GB
Needs at Q4
41.8 GB
Headroom
86.2 GB
Speed (Q4_K_M)
~3.9 tok/s

Every quantization

Qwen2.5 72B (72B) on 128 GB

QuantWeightsNeedsFitEst. speed
Q4_K_M40 GB41.8 GBFits easily~3.9 tok/s
Q5_K_M48 GB49.8 GBFits easily~3.3 tok/s
Q8_078 GB79.8 GBFits~2 tok/s
FP16144 GB145.8 GBToo big -

Estimates. Needs = weights + 1.8 GB for KV cache and runtime at modest context. Speed is single-stream decode scaled from the DGX Spark (GB10)'s 273.2 GB/s bandwidth (Estimated from memory bandwidth (llama.cpp decode)). Long contexts need more memory and run slower.

Run it

Copy, paste, go

ollama run qwen2.5:72b
llama-server -hf Qwen/Qwen2.5-72B-Instruct-GGUF -ngl 99

Ollama pulls a Q4 build by default. Pick a larger quant only if the table above says it fits. Full Qwen2.5 72B details →

Compare

Same model, other hardware - and other models on this card

Qwen2.5 72B on other hardware

Also runs on the DGX Spark (GB10)

Frequently asked

Quick answers

Can the NVIDIA DGX Spark (GB10) run Qwen2.5 72B?

Yes. Qwen2.5 72B fits the DGX Spark (GB10)'s 128 GB at Q4_K_M (about 41.8 GB needed), running at an estimated 3.9 tok/s. There's room for up to Q8_0 if you want higher quality.

How much VRAM does Qwen2.5 72B need?

About 40 GB of weights at Q4_K_M, 78 GB at Q8_0 and 144 GB at FP16. Add roughly 1.8 GB for the KV cache and runtime at modest context lengths.

How fast is Qwen2.5 72B on the DGX Spark (GB10)?

Roughly 3.9 tokens per second at Q4_K_M (single user, llama.cpp-style decode). This is an estimate from the DGX Spark (GB10)'s 273.2 GB/s memory bandwidth.