Home › Models › Nex-N2-Pro › NVIDIA DGX Spark (GB10)
Nex-N2-Pro on the NVIDIA DGX Spark (GB10)
Nex-N2-Pro runs at about 18.9 tokens/s on the NVIDIA DGX Spark (GB10) (measured, IQ1_M, CUDA, single-stream llama-server decode).
IQ1_M · CUDA
The source gives no prompt-processing figure.
For comparison, our size-scaling estimate for this combination is 37.8 tok/s.
Every run
| Quant | Backend | Generate tok/s | Prompt tok/s | Test | Build / by / notes | Source |
|---|---|---|---|---|---|---|
| IQ1_M | CUDA | 18.9 | – | decode per session, llama-server, 1 concurrent request, 1024-token prompt | Single DGX Spark, whole 91 GB IQ1_M model in memory, flash attention on, q8_0 KV cache, no speculative decoding; one request at a time (concurrency 1 point of a load sweep). llama.cpp build not stated by the source. | jvr0x/dgx-spark-bench results (GitHub) |
tg128 = generating 128 tokens; pp512 = reading a 512-token prompt (llama-bench). LocalScore averages several prompt and output lengths, so its numbers run lower than tg128. "decode, llama-server" = single-stream generation speed measured against a running llama-server (no speculative decoding).
No card yet?
Try it on a rented GPU (RunPod)Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.
All results for the NVIDIA DGX Spark (GB10) · Nex-N2-Pro on other hardware
Embed this result
HTML
<a href="https://rigbench.pages.dev/model/nex-n2-pro/on/nvidia-dgx-spark-gb10/"><img src="https://rigbench.pages.dev/badge/nex-n2-pro--nvidia-dgx-spark-gb10.svg" alt="Nex-N2-Pro · NVIDIA DGX Spark (GB10): 18.9 tok/s (IQ1_M, measured)"></a>
Markdown (GitHub README, forums)
[](https://rigbench.pages.dev/model/nex-n2-pro/on/nvidia-dgx-spark-gb10/)