community pool / model answer

Qwopus3.5-27B-v3

chat · Jackrong/Qwopus3.5-27B-v3

12.9 tok/s · model median · 23 runs in the model record.

Reference rig	Quantization	Median result	Basis	TTFT (ms)	Peak VRAM	Max context	Engines
RTX 3090 24GB	4-bit	25.8 tok/s	measured	7,951	0 GB	131,072	llama.cpp
Ryzen AI Max 395 128GB	4-bit	12.9 tok/s	measured	-	17.5 GB	32,768	llama.cpp
Ryzen AI Max 395 128GB	5-bit	11.3 tok/s	measured	-	19.9 GB	32,768	llama.cpp
Ryzen AI Max 395 128GB	6-bit	9.9 tok/s	reported	-	23.5 GB	32,768	llama.cpp
Ryzen AI Max 395 128GB	8-bit	7.9 tok/s	reported	-	29.2 GB	32,768	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.