Qwen

Qwen

Qwen/Qwen3.6-35B-A3B

Quantizations

Quant	Quantized by	Size	Decode	Prefill	Score	Actions
ud-iq1_m	Unsloth Unsloth	9.4 GB	N/A	N/A	N/A	Run
ud-iq2_xxs	Unsloth Unsloth	10.0 GB	N/A	N/A	N/A	Run
ud-iq2_m	Unsloth Unsloth	10.7 GB	N/A	N/A	N/A	Run
3bit	Frank Denis Frank Denis	14.1 GB	N/A	N/A	N/A	Run
iq4_xs	DuoNeural DuoNeural	17.4 GB	73.5 tok/s	1,504.0 tok/s	Runs well	Run
nvfp4	MLX Community MLX Community	19.0 GB	95.8 tok/s	2,360.1 tok/s	Runs well	Run
4bit	MLX Community MLX Community	19.0 GB	90.2 tok/s	2,150.3 tok/s	Runs well	Run
4bit-dwq	MLX Community MLX Community	19.2 GB	N/A	N/A	N/A	Run
ud-q4_k_s	Unsloth Unsloth	19.5 GB	N/A	N/A	N/A	Run
q4_k_m	LM Studio LM Studio	19.7 GB	N/A	N/A	N/A	Run
q4_k_m	DuoNeural DuoNeural	19.7 GB	71.5 tok/s	1,532.6 tok/s	Runs well	Run
6bit	LM Studio LM Studio	27.1 GB	N/A	N/A	N/A	Run
8bit	MLX Community MLX Community	35.1 GB	N/A	N/A	N/A	Run
ud-q8_k_xl	Unsloth Unsloth	35.8 GB	N/A	N/A	N/A	Run

Device Comparison

Results include trials with 4,096 input tokens and 1,024 output tokens only.