readygpu

ASUS Ascent GX10 128GB vs Beelink GTR9 Pro Ryzen AI Max+ 395 128GB

For running models locally, priced in the US. At 32k context the first runs 73 models fully in memory, the second 72.

In short

Both have 128 GB of memory for models; the ASUS Ascent GX10 128GB has 273 GB/s of bandwidth and the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB 256 GB/s. At 32k context the ASUS Ascent GX10 128GB runs 73 of the models we track fully in memory and the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB 72; only the ASUS Ascent GX10 128GB holds GLM 5.3 Flash. On the models both run fully in memory, the ASUS Ascent GX10 128GB is about 1.2× as fast as the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB (estimated, median across models).

Updated . Speeds are estimates; how they are worked out.

ASUS Ascent GX10 128GB

$6,100

eBay

Add to build
Beelink GTR9 Pro Ryzen AI Max+ 395 128GB

Not priced here

Add to build

Specifications

SpecASUS Ascent GX10 128GBBeelink GTR9 Pro Ryzen AI Max+ 395 128GB
ChipGB10 Grace BlackwellRyzen AI Max+ 395 (Radeon 8060S)
Memory bus256-bit256-bit
Memory bandwidth273 GB/s256 GB/s
FP16 compute62 TFLOPS59.4 TFLOPS
SoftwareCUDAROCm
Unified memory128 GB128 GB
GPU can use by default119 GB96 GB
GPU can use, tuned122 GB120 GB
Operating systemdgx-oswindows
Processor20-core Arm (GB10)Ryzen AI Max+ 395
Storage1000 GB2000 GB

What each runs

Estimated decode speed with the most precise quant that fits each one est.

ModelASUS Ascent GX10 128GBBeelink GTR9 Pro Ryzen AI Max+ 395 128GB
DeepSeek V4 Flash11tok/sUD-Q2_K_XL9tok/sUD-Q2_K_XL
GLM 5.3 Flash11tok/sUD-Q2_K_XLwon't run
Qwen 3.8 Flash Next2tok/sQ4_K_M3tok/sUD-Q3_K_XL
MiniMax M2.78tok/sUD-Q3_K_M7tok/sUD-Q2_K_XL
Qwen 3.8 27B5tok/sQ8_05tok/sQ8_0
Qwen 3.5 122B-A10B11tok/sQ6_K9tok/sQ6_K
Qwen 3.6 27B6tok/sQ8_05tok/sQ8_0
gpt-oss 120B23tok/sMXFP420tok/sMXFP4
Qwen3 Coder Next20tok/sQ8_017tok/sQ8_0
Nemotron 3 Super 120B-A12B9tok/sUD-Q6_K9tok/sQ5_K_M
Gemma 4 31B5tok/sQ8_04tok/sQ8_0
Mistral Small 4 119B15tok/sUD-Q6_K13tok/sUD-Q6_K
Qwen 3.6 35B-A3B27tok/sQ8_022tok/sQ8_0
Qwen 3.5 27B6tok/sQ8_05tok/sQ8_0
Gemma 4 26B-A4B20tok/sQ8_017tok/sQ8_0
GLM 4.5 Air4tok/sQ8_04tok/sQ5_K_M
Qwen 3.5 35B-A3B27tok/sQ8_022tok/sQ8_0
Qwen3 Next 80B-A3B25tok/sQ8_021tok/sQ8_0
GLM 4.7 Flash14tok/sQ8_012tok/sQ8_0
Devstral Small 2 24B6tok/sQ8_05tok/sQ8_0
Nemotron 3.5 Lightning 30B-A3B28tok/sUD-Q6_K_XL24tok/sUD-Q6_K_XL
gpt-oss 20B31tok/sMXFP426tok/sMXFP4
Kimi Linear 48B-A3B29tok/sQ8_024tok/sQ8_0
Qwen3 Coder 30B-A3B16tok/sQ8_013tok/sQ8_0
Qwen3 32B4tok/sQ8_03tok/sQ8_0
Gemma 4 12B13tok/sQ8_010tok/sQ8_0
Qwen 3.5 9B16tok/sQ8_013tok/sQ8_0
Mistral Small 3.2 24B6tok/sQ8_05tok/sQ8_0
Gemma 3 27B5tok/sQ8_04tok/sQ8_0
DeepSeek R1 Distill Qwen 32B4tok/sQ8_03tok/sQ8_0
Llama 3.3 70B2tok/sQ8_02tok/sQ8_0
Ministral 3 14B9tok/sQ8_07tok/sQ8_0
Qwen3 14B8tok/sQ8_07tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B13tok/sQ8_010tok/sQ8_0
Qwen3 8B13tok/sQ8_010tok/sQ8_0
Phi-4 14B8tok/sQ8_06tok/sQ8_0
Qwen 3.5 4B30tok/sQ8_025tok/sQ8_0
Ministral 3 8B13tok/sQ8_010tok/sQ8_0
Granite 4.2 8B12tok/sQ8_010tok/sQ8_0
Gemma 4 E4B19tok/sQ8_015tok/sQ8_0
LFM2.5 8B-A1B38tok/sQ8_032tok/sQ8_0
Llama 3.1 8B16tok/sQ6_K13tok/sQ6_K
Qwen2.5 0.5B181tok/sQ8_0151tok/sQ8_0
Qwen3.5 0.8B137tok/sQ8_0114tok/sQ8_0
Qwen3.5 2B91tok/sQ5_K_M76tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B60tok/sQ8_050tok/sQ8_0
MiniCPM5 2B41tok/sQ8_034tok/sQ8_0
LFM2.5 2.6B35tok/sQ8_029tok/sQ8_0
LFM2.5 VL 3B35tok/sQ8_029tok/sQ8_0
Mellum2.1 12B A2.5B Thinking32tok/sQ8_027tok/sQ8_0
granite vision 4.1 4b27tok/sQ8_022tok/sQ8_0
granite 4.2 3b26tok/sQ8_021tok/sQ8_0
Kimi VL A3B Thinking 250625tok/sQ8_021tok/sQ8_0
Kimi VL A3B24tok/smlx-8bit20tok/smlx-8bit
LFM2.5 24B A2B21tok/sQ8_018tok/sQ8_0
GLM 4 9B 041418tok/smlx-5bit15tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B17tok/sQ8_014tok/sQ8_0
GLM 4.6V Flash15tok/sQ8_012tok/sQ8_0
DeepSeek R1 Distill Llama 8B13tok/sQ8_011tok/sQ8_0
GLM 4.1V 9B Thinking13tok/smlx-8bit11tok/smlx-8bit
Ministral 3 8B Reasoning 251213tok/sQ8_010tok/sQ8_0
DeepSeek R1 Distill Qwen 14B8tok/sQ8_06tok/sQ8_0
Devstral Small 25076tok/sQ8_05tok/sQ8_0
Magistral Small 25066tok/sQ8_05tok/sQ8_0
Magistral Small 25076tok/sQ8_05tok/sQ8_0
Magistral Small 25096tok/sQ8_05tok/sQ8_0
GLM 4.5V5tok/sQ8_06tok/sQ4_K_M
GLM 4.6V5tok/sQ8_05tok/sQ4_K_M
granite 4.2 30b4tok/sQ8_04tok/sQ8_0
DeepSeek R1 Distill Llama 70B2tok/sQ8_02tok/sQ8_0
Kimi Dev 72B2tok/sQ8_02tok/sQ8_0
Devstral 2 123B Instruct 25122tok/sQ6_K1tok/sQ5_K_M
Mistral Medium 3.5 128B2tok/sQ6_K1tok/sQ5_K_M
DeepSeek R1won't runwon't run
DeepSeek R1 0528won't runwon't run
DeepSeek R1 Zerowon't runwon't run
DeepSeek V3 0324won't runwon't run
DeepSeek V3.1won't runwon't run
DeepSeek V3.1 Terminuswon't runwon't run
DeepSeek V3.2won't runwon't run
DeepSeek V4 Flash 0731won't runwon't run
DeepSeek V4 Flash Vision Expwon't runwon't run
GLM 4.6won't runwon't run
GLM 4.7won't runwon't run
GLM 5.1won't runwon't run
GLM 5.2won't runwon't run
GLM 5.3won't runwon't run
Kimi K2won't runwon't run
Kimi K2 Instruct 0905won't runwon't run
Kimi K2 Thinkingwon't runwon't run
Kimi K2.5won't runwon't run
MiniMax M2won't runwon't run
MiniMax M2.1won't runwon't run
MiniMax M2.5won't runwon't run
MiniMax M3won't runwon't run

Questions

Which is better for local AI, the ASUS Ascent GX10 128GB or the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB?
Choose the ASUS Ascent GX10 128GB to run bigger models. Choose the ASUS Ascent GX10 128GB for speed.
Which can run bigger models, the ASUS Ascent GX10 128GB or the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB?
At 32k context the ASUS Ascent GX10 128GB runs 73 of the models we track fully in memory and the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB 72; only the ASUS Ascent GX10 128GB holds GLM 5.3 Flash.
Which is faster for local AI, the ASUS Ascent GX10 128GB or the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB?
On the models both run fully in memory, the ASUS Ascent GX10 128GB is about 1.2× as fast as the Beelink GTR9 Pro Ryzen AI Max+ 395 128GB (estimated, median across models).

More comparisons