DiggTop

Qwen 3.8 Flash

Qwen family · 125.0B params

8recorded runs
86.8best tok/s
8verified
Fastest GPUs for this model (best tok/s)
  1. DGX Spark86.8

All benchmarks Methodology

GPUModelQuantBackendCtx tok/sprompt tok/sVRAM GBSource
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd3204886.8 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd360071.7 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd36400067.4 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd311000066.0 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd33500058.0 980.0src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd324000055.8 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd33200055.6 src
DGX Spark · 128 GBQwen 3.8 FlashEXL3 3.05bpwexllamav2 523ecd3800051.5 src