COMPUTER RECORD · SOURCED SPECIFICATIONS
ASUS ROG NUC 16: specifications and local LLMs
64 GB system RAM and 896 GB/s rated memory bandwidth. 64GB system RAM and NVIDIA CUDA with 16GB dedicated GDDR7 VRAM.
CPU, GPU, memory and storage
- CPU
- Core Ultra 9 290HX Plus · 24 cores / 24 threads
- GPU
- RTX 5080 Laptop GPU · 16GB GDDR7
- Installed memory
- 64 GB · DDR5-5600 system RAM; separate GDDR7 VRAM · Owner-supplied Amazon.com screenshot
- GPU memory pool
- 16 GB
- Memory bandwidth
- 896 GB/s rated / theoretical · NVIDIA RTX 5080 Laptop GPU rated bandwidth; this describes dedicated VRAM, not DDR5 system RAM.
- Storage
- 2000 GB · Owner-supplied Amazon.com screenshot
- Operating system
- Windows 11 (screenshot)
- Expansion
- Model family: SO-DIMM; M.2; Thunderbolt 4
- NPU advertised peak
- 13 TOPS · not LLM tokens/s
- Whole-chip advertised peak
- 36 TOPS · distinct from NPU-only performance
- FP4 advertised peak
- 1334 TFLOPS · not measured LLM throughput
Only the 16GB VRAM pool is local to the GPU; system RAM does not add to dedicated VRAM.
The screenshot selects the white 2026 Ultra 9 / RTX 5080 / 64GB / 2TB model. Manufacturer sources identify the RTX 5080 as the Laptop GPU.
Owner-supplied Amazon.com screenshot reviewed 24 Sep 2026 and matched by the owner to ASIN B0HB1PVWDK. The screenshot shows the product title and price, but not the ASIN or capture time. It is not an independent live listing check. Current seller, stock and checkout price can change.
What local models could ASUS ROG NUC 16 run?
The 16 GB dedicated-VRAM pool gives a 12.8 GB initial weight budget using our 20% (minimum 1 GB) reserve. The 64 GB system RAM is a separate pool and is not added to VRAM.
86 catalog files pass this size check. The largest is Winnow 12B at 12.67 GB (Q8_0). This is a catalog weight limit, not a measured maximum practical model size. Context length, runtime and architecture support can rule out a candidate.
Model snapshot: 14 October 2026. The examples below use MMLU-Pro; models without that score are not ranked in this table.
| Model / quantization | Weights | Parameters | Score / 100 | Score / GB | Budget left | Sources |
|---|---|---|---|---|---|---|
| MiniCPM5 2B | 1.561 GB | 2.517B | 70.8 | 45.346 | 11.239 GB | Get weights Score source |
| Qwen3.5-4B | 2.741 GB | 4.206B | 79.1 | 28.859 | 10.059 GB | Get weights Score source |
| LFM2.5 230M | 0.153 GB | 0.23B | 20.25 | 132.002 | 12.647 GB | Get weights Score source |
| Qwen3.5-9B | 5.681 GB | 9.653B | 82.5 | 14.523 | 7.119 GB | Get weights Score source |
| gemma-4-12B-it | 7.122 GB | 11.96B | 77.2 | 10.84 | 5.678 GB | Get weights Score source |
Choose this computer in the memory tool · Compare its memory tier
Published LLM performance
Local AI Stats has not tested this computer. No matching token-speed result is recorded for this configuration.
Advertised compute peaks and bandwidth are not tokens per second. Compare the same model, quantization, context and concurrency.
Price and value
ASIN: B0HB1PVWDK · Amazon.com listing
$4,999 USD reference. Amazon.com price shown in an owner-supplied screenshot reviewed 24 Sep 2026; capture time not provided. Excludes shipping, import charges, tax and optional discounts. This is a dated reference, not a live checkout quote.
Reference price per installed-memory GB: $78.109. Per dedicated-VRAM GB: $312.438. This measures cost per GB, not speed or total running cost.
As an Amazon Associate I earn from qualifying purchases. Check the exact configuration, seller, marketplace and current price before buying.
GET NOW · Amazon