TokenAssemble

Can an NVIDIA DGX Spark run Qwen2.5-3B-Instruct?

Yes. Qwen2.5-3B-Instruct needs 4.01GB at Q8_0 with llama.cpp and 8K context; the NVIDIA DGX Spark has 128GB of unified memory (~90% GPU-addressable) (114.7GB usable) — grade A+, with decode speed in the interactive tier.

NVIDIA DGX Spark · Qwen2.5-3B-Instruct · Q8_0

Fits — grade A+

InteractivebetaNew hardware class — tier not yet calibrated
weights 3.29 GBKV cache 0.30 GBoverhead 0.42 GBpool 114.7 GBneeds 4.01 GB

assumes llama.cpp · batch 1 · 8K context · fp16 KV cache

$4,699 MSRP · as of 2026-07-11

Affiliate link — we may earn a commission. Verdicts are computed before any link is attached.

Every quant, graded

NVIDIA DGX Spark × Qwen2.5-3B-Instruct, llama.cpp, 8K context.

QuantFile sizeFitSpeed tier
Q8_03.3 GBFits · A+interactive · beta
Q6_K2.5 GBFits · A+interactive · beta
Q5_K_M2.2 GBFits · A+interactive · beta
Q4_K_M1.9 GBFits · A+interactive · beta

Qwen2.5-3B-Instruct on similar hardware

More models on the NVIDIA DGX Spark

Verdict computed from published specs and measured GGUF file sizes — see how the math works.