Hardware Corner Qwen3.8-27B Q4_K_S: NVIDIA GB10 (Dell Pro Max / DGX Spark) 128GB unified tg 5.94 t/s @256k ctx
5.94 tok/s per-request decode (batch 1, 256k context)
Fields
id
benchmarks/hc38-nvidia-gb10-qwen38-27b-tg-256kMetric
decode_tok_s
Value
5.94
Unit
tok/s per-request decode (batch 1, 256k context)
Model
Qwen3.8 27B Q4_K Small GGUF
Methodology
THIRD-PARTY MEASURED (hardware-corner.net 'We Tested Qwen3.8 27B', accessed 2026-10-15). llama.cpp build 153d324bc (10364), Ubuntu 24.04, NVIDIA driver 595.58.03, CUDA 12.8, Qwen3.8-27B Q4_K Small GGUF (16.68 GiB, 27.32B params; llama.cpp reports arch 'qwen35'), Flash Attention on, MTP off, full GPU offload. NVIDIA GB10 (Dell Pro Max / DGX Spark) 128GB unified. pp = prompt-processing tok/s, tg = token-generation tok/s at stated prefilled context. Per-request batch-1.
Measured by
third_party
Reproducible
yes
notes
256k context - only the 128 GB unified part in this test could hold it.
Notes
256k context - only the 128 GB unified part in this test could hold it.