Qwen3-VL-8B-Instruct

16 records across 4 hardware.

hardware	model	quant	backend	tok/s	pp tok/s	W	ttft s	ctx/tg	date	class	flags	source
M3 Ultra	Qwen3-VL-8B-Instruct	4bit	oMLX	90.5	1044.0	–	–	8192	2026-10-15	community	–	source
M3 Ultra	Qwen3-VL-8B-Instruct	4bit	oMLX	72.7	930.5	–	–	16384	2026-10-15	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	62.8	2876.0	–	–	1024	2026-10-11	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	60.4	3059.0	–	–	4096	2026-10-11	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	57.0	3108.0	–	–	1024	2026-10-11	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	57.0	3108.0	–	–	1024	2026-10-11	community	–	source
M4 Pro	Qwen3-VL-8B-Instruct	4bit	oMLX	51.7	467.8	–	–	1024	2026-10-15	community	–	source
M4 Pro	Qwen3-VL-8B-Instruct	4bit	oMLX	47.3	444.8	–	–	4096	2026-10-15	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	21.3	224.6	–	–	1024	2026-03-09	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	21.3	224.6	–	–	1024	2026-03-09	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	20.5	158.6	–	–	1024	2026-10-12	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	18.9	217.1	–	–	4096	2026-03-09	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	18.9	217.1	–	–	4096	2026-03-09	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	18.5	815.1	–	–	131072	2026-10-11	community	–	source
M5 Max	Qwen3-VL-8B-Instruct	8bit	oMLX	18.5	815.1	–	–	131072	2026-10-11	community	–	source
M4	Qwen3-VL-8B-Instruct	4bit	oMLX	18.4	152.8	–	–	4096	2026-10-12	community	–	source