PRE-ORDER
NIMO lab results on AMD Ryzen™ AI Max+ PRO 495, Radeon™ 8065S, 192GB unified memory and NimoOS. All results are in tokens/s.
| Model / quantization | Total / active parameters | Model size (GiB) | Engine / backend | Speculative decoding | 16K prefill | Short decode | 16K decode |
|---|---|---|---|---|---|---|---|
| GLM-5.3-Flash / UD-IQ4_XS | 320B / 18B | 146.05 | Unsloth llama.cpp / HIP 7.2.4 | Off | 162.95 | 12.26 | 11.45 |
| GLM-5.3-Flash / UD-IQ4_XS | 320B / 18B | 146.05 | Unsloth llama.cpp / HIP 7.2.4 | MTP 2 | 162.97 | 19.60 | 16.75 |
| DeepSeek-V4-Flash-0731 / UD-Q8_K_XL | 284B / 13B | 150.75 | llama.cpp / HIP 7.2.4 | Off | 153.68 | 8.87 | 8.47 |
| DeepSeek-V4-Flash-0731 / Q4_0_ROCMFP4_STRIX | 284B / 13B | 140.98 | ROCmFPX / HIP 7.2.4 | Off | 77.50 | 10.59 | 7.81 |
| DeepSeek-V4-Flash-0731 / Q4_0_ROCMFP4_STRIX | 284B / 13B | 140.98 | ROCmFPX / Vulkan | Off | 75.11 | 10.25 | 8.53 |
| DeepSeek-V4-Flash-0731 / Q4_0_ROCMFP4_STRIX | 284B / 13B | 140.98 | ROCmFPX / HIP 7.2.4 | DSpark 4 | 76.58 | 17.03 | 12.58 |
| DeepSeek-V4-Flash-0731 / antirez Q4KExperts mixed quantization | 284B / 13B | 153.33 | Gufo / HIP 7.2.4 | Off | 64.22 | 14.53 | 13.19 |
| DeepSeek-V4-Flash-0731 / antirez Q4KExperts mixed quantization | 284B / 13B | 153.33 | Gufo / HIP 7.2.4 | DSpark 5 | 64.29 | 16.83 | 28.29 |
NIMO lab report, November 2, 2026. Single GPU, concurrency 1, greedy decoding, three valid runs per configuration. Prefill uses 16,384 input tokens. Decode generates 512 tokens. The short prefix is 16 tokens for Gufo and 512 for the other engines. Gufo uses synthetic inputs, while the other rows use fixed text. GLM decode reuses the prompt cache and MTP changes the output. Gufo reached 28.29 tokens/s on the synthetic 16K sample with 100% draft acceptance. Model size is on-disk size. These workloads do not support a like-for-like engine ranking.
696K YouTube subscribers
Quote: This video shows the Nimo 17.3" Ryzen AI 9 HX 370 delivering a fast 12-core CPU, sleek elegant design, and easy upgradability, while proving that dual-channel memory nearly doubles gaming performance on the Radeon 890M.
Product Used：Nimo 17.3" AMD Ryzen AI 9 HX 370 Laptop
415K YouTube subscribers
Quote: This video reviews the Nimo AI 370 , highlighting its AMD Ryzen AI 9 HX 370 processor, 144Hz 17.3-inch display, customizable RAM and storage options, all-day battery life, and USA-based warranty and repair support.
Product Used：Nimo 17.3" AMD Ryzen AI 9 HX 370 Laptop
1.17M YouTube subscribers
Quote: This video shows the NIMO N157 handling heavy Photoshop edits and 3D rendering without slowdown, thanks to its Ryzen 9 and Radeon 780M, while its bright 17.3-inch display stays clear even in full sunlight.
Product Used: Nimo N157 Laptop
415K YouTube subscribers
Quote: This video shows a live, unscripted teardown of the NIMO AI Mini PC, highlighting its AMD Ryzen AI Max+ 395 processor, 128GB RAM, rugged metal build, and versatility for gaming, AI, and creative work.
Product Used ：NIMO AI Mini PC 4L version
1.14M YouTube subscribers
Quote：This video showcases the NIMO N175's crisp 17.3-inch display and smooth AMD Ryzen 9 performance, running games and high-resolution video without lag, earning a 9.5 out of 10 rating for everyday productivity.
Product Used：NIMO N175 Laptop
