Microsoft Surface Laptop Ultra (RTX Spark N1X): llama.cpp speeds on 2026 models (Micro Center)
| GPU(s) | Nvidia RTX Spark N1X, 6,144-core Blackwell GPU (unified) |
|---|---|
| VRAM / unified memory (GB) | |
| CPU | Nvidia RTX Spark N1X, 20-core Grace CPU Amazon · eBay (used) · Newegg |
| Motherboard | |
| RAM | |
| PSU | |
| Other | |
| Reported total cost | |
| Power | Avg GPU power while generating: 43-48 W |
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s |
|---|---|---|---|
| Nemotron 3 Nano Omni 30B | general, short (~90-token) prompt | llama.cpp (Nvidia ARM64 build) | 63.9 |
| GLM 4.7 Flash | general, short prompt | llama.cpp (Nvidia ARM64 build) | 55.6 |
| Qwen3.5 122B | A10B, general, short prompt | llama.cpp (Nvidia ARM64 build) | 35.2 |
| Muse Glimmer 30B | general, short prompt | llama.cpp (Nvidia ARM64 build) | 15.9 |
| Qwen3.8 27B | general, short prompt | llama.cpp (Nvidia ARM64 build) | 14.5 |
Micro Center benchmarks, 15 Oct 2026. Long (~9,000-token) prompts were within a few tok/s (e.g. Qwen3.8 27B 14.1, Qwen3.5 122B 34.0). The article does not state the tested unit's memory configuration or quantization.
Source: https://www.microcenter.com/site/mc-news/article/benchmarking-the-rtx-spark.aspx
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: RunPod
Quick answers: Can RTX Spark N1X run GLM 4.7 Flash? · Can RTX Spark N1X run Muse Glimmer 30B? · Can RTX Spark N1X run Nemotron 3 Nano Omni 30B? · Can RTX Spark N1X run Qwen3.5 122B? · Can RTX Spark N1X run Qwen3.8 27B?