community pool / model answer

Qwen3.6-27B-DFlash

chat · z-lab/Qwen3.6-27B-DFlash

39.2 tok/s · model median · 10 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
Arc Pro B70 32GB ×2	4-bit	38.2 tok/s	measured	-	- GB	512	llama.cpp
Arc Pro B70 32GB ×3	4-bit	41.4 tok/s	measured	-	- GB	512	llama.cpp
Arc Pro B70 32GB ×4	4-bit	31.5 tok/s	reported	-	- GB	512	llama.cpp
GB10 Grace Blackwell 128GB	8-bit	62.1 tok/s	reported	1	- GB	2,048	vllm
RTX 5090 32GB	4-bit	215.4 tok/s	reported	470	23 GB	1,024	llama.cpp

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.