DiggTop

CMP 170HX vs DGX Spark x2

Best verified output tok/s per model (single-stream generation). Hover a value for the quant.

CMP 170HX16 verified runs · best 255.9 tok/s
DGX Spark x210 verified runs · best 474.0 tok/s
ModelCMP 170HXDGX Spark x2
Qwen3.8-Flash-Next-hibrid48474.0 NVFP4 (output head)
Qwen3.8-27B255.9 W4A16
NVIDIA-Nemotron-3.5-Lightning-30B-A3B165.1 Q8_0
DeepSeek-V4.1-Flash154.0 EXL3 2.9bpw
Qwen3.6-35B-A3B87.2 UD-Q8_K_XL
Ornith-1.5-35B-A3B87.1 Q8_0
Qwen3.8-Flash-Next46.1 UD-IQ1_S80.0 NVFP4
Ornith-1.5-9B61.5 BF16
Qwen3.8-27B-MTP46.4 INT8-W8A16
GLM-5.3-Flash46.3 NVFP4
Llama-4-Scout-17B-16E-Instruct44.5 UD-Q3_K_XL
gemma-4-31B-it28.3 NVFP4
granite-4.2-30b23.7 Q8_0
Llama-3.3-70B-Instruct20.1 NVFP4

Head-to-head (models tested on both): CMP 170HX 0 · DGX Spark x2 1 · tied 0 · data as of Sep 23, 2026. Full rules: methodology.

Related reading: RTX 5090 vs 3090 · quantization · Value index · Full board