// RUN — FITS

Can the NVIDIA DGX Spark run Qwen2.5-Math-7B-Base?

Yes — here's how fast and up to what context, with real measured numbers.

Qwen2.5-Math-7B-Base on the NVIDIA DGX Spark

QuantFits?Max contextSpeedBeyond context
FP16 yes 761k 14.2-17 t/s ~0.5-0.6 (slow)
Q8 yes 814k 28.5-33.9 t/s ~1-1.2 (slow)
Q4 yes 838k 50.8-60.6 t/s ~1.9-2.2 (slow)

Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.

→ All models on the NVIDIA DGX Spark → Qwen2.5-Math-7B-Base specs → Other hardware

FAQ

Can the NVIDIA DGX Spark run Qwen2.5-Math-7B-Base?▶

Yes. At Q4 it generates ~50.8-60.6 tok/s and fits up to 838k context; beyond that it spills to SSD swap and slows to ~1.9-2.2 tok/s.

How much context fits?▶

At Q4, up to about 838k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.