community pool / model answer

Qwen3-Coder-Next

code · Qwen/Qwen3-Coder-Next

55.8 tok/s · model median · 16 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	58.5 tok/s	reported	-	49 GB	2,048	llama.cpp
Arc Pro B70 32GB ×4	8-bit	71.7 tok/s	reported	95	127.2 GB	262,144	vllm
Radeon AI Pro R9700 32GB ×2	4-bit	80.8 tok/s	reported	250	- GB	717	llama.cpp
Radeon AI Pro R9700 32GB ×3	5-bit	64.5 tok/s	measured	432	70 GB	786	llama.cpp
Ryzen AI Max 395 128GB	4-bit	55.8 tok/s	measured	-	46 GB	80,000	llama.cpp
Ryzen AI Max 395 128GB	5-bit	51.2 tok/s	measured	1,223	53.7 GB	80,000	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.