Builds / Nvidia RTX Spark N1X, 6,144-core Blackwell GPU (unified)

RTX Spark N1X (128 GB): Qwen3.8 Flash-Next 177B-A3B UD-IQ4_XS at 31.0 tok/s, Windows on Arm (llama-bench)

Last updated: 2 Nov 2026

Source: github.com ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source

GPU(s): Nvidia RTX Spark N1X, 6,144-core Blackwell GPU (unified)
VRAM / unified memory (GB): 128
CPU: Nvidia RTX Spark N1X
Amazon · eBay (used) · Newegg
Motherboard: not stated
RAM: 128 GB unified memory
PSU: not stated
Other: Windows ARM64, llama.cpp CUDA backend build 12024 (ddf48f85c); flash attention on, KV cache q8_0/q8_0, -ngl 999, -b 2048 -ub 512, split mode none, 5 repetitions
Reported total cost: not stated
Power: not stated

Measured speed (generation tokens/sec)

Model	Quant	Engine	tok/s	Prompt processing (tok/s)	Details
Qwen3.8 Flash-Next	UD-IQ4_XS (4.25 bpw, 87.24 GiB)	llama.cpp build 12024 (CUDA, Windows ARM64)	30.97	588 @ 8,192-token prompt est. TTFT @8k ≈ 13.9 s (calculated from measured prefill)	context: llama-bench tg512 at depth 0📏 llama-bench tg512 30.97 ±0.05 (tg128 30.83 ±0.32), 5 repetitions, FA on, q8_0 KV

Prompt processing = prefill speed (how fast the prompt is read) as reported by the source, with the prompt length it was measured at; blank = the source does not report it. est. TTFT @8k is calculated from the measured prefill speed (8,192 ÷ prefill tok/s), shown only when prefill was measured on a prompt of 512+ tokens. It is not a measurement and ignores load time and caching. measured TTFT is the source's own time-to-first-token. Why prefill matters for coding assistants.

llama.cpp GitHub discussion by knmun, 1 Nov 2026. Whole 87 GiB model on the laptop chip's GPU. The laptop model is not named. pp2048 was unstable across runs (313.36 ±162.68, range 167.61-592.86), so only pp8192 is listed.

Source: https://github.com/ggml-org/llama.cpp/discussions/31494

Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.

Embed this result

Copy into a README or blog. Links back to this build. See badge docs (CC BY 4.0 attribution). The SVG does not invent speeds — numbers stay on the build page.

Markdown

[![Measured on RigBase](https://getrigbase.com/badge/rtx-spark-n1x-128gb-qwen38-flash-next-iq4xs-windows-arm64.svg)](https://getrigbase.com/builds/rtx-spark-n1x-128gb-qwen38-flash-next-iq4xs-windows-arm64/)

HTML

<a href="https://getrigbase.com/builds/rtx-spark-n1x-128gb-qwen38-flash-next-iq4xs-windows-arm64/" rel="noopener"><img src="https://getrigbase.com/badge/rtx-spark-n1x-128gb-qwen38-flash-next-iq4xs-windows-arm64.svg" alt="Measured on RigBase" height="20"/></a>

Preview: · How to use

Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)

Quick answers: Can RTX Spark N1X run Qwen3.8 Flash-Next?