Can the DGX Spark (GB10) run Nemotron 3.5 Lightning?

Yes. Nemotron 3.5 Lightning fits the DGX Spark (GB10)'s 128 GB at Q4_K_M (about 11.8 GB needed), running at an estimated 16 tok/s. There's room for up to FP16 if you want higher quality.

VRAM
128 GB
Needs at Q4
11.8 GB
Headroom
116.2 GB
Speed (Q4_K_M)
~16 tok/s

Every quantization

Nemotron 3.5 Lightning (17.8B) on 128 GB

QuantWeightsNeedsFitEst. speed
Q4_K_M10 GB11.8 GBFits easily~16 tok/s
Q5_K_M13 GB14.8 GBFits easily~12 tok/s
Q8_019 GB20.8 GBFits easily~8.3 tok/s
FP1636 GB37.8 GBFits easily~4.4 tok/s

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.

Compare

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

Nemotron 3.5 Lightning on other hardware

Also runs on the DGX Spark (GB10)

Frequently asked

Quick answers

Can the NVIDIA DGX Spark (GB10) run Nemotron 3.5 Lightning?

Yes. Nemotron 3.5 Lightning fits the DGX Spark (GB10)'s 128 GB at Q4_K_M (about 11.8 GB needed), running at an estimated 16 tok/s. There's room for up to FP16 if you want higher quality.

How much VRAM does Nemotron 3.5 Lightning need?

About 10 GB of weights at Q4_K_M, 19 GB at Q8_0 and 36 GB at FP16. Add roughly 1.8 GB for the KV cache and runtime at modest context lengths.

How fast is Nemotron 3.5 Lightning on the DGX Spark (GB10)?

Roughly 16 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.