Qwen3 14B

101 records across 28 hardware.

hardware	model	quant	backend	tok/s	pp tok/s	W	ttft s	ctx/tg	date	class	flags	source
RTX 5090	Qwen3 14B 14B	Q4_K	llama.cpp	123.8	6497.6	–	–	4096	–	sourced	–	source
RTX 5090	Qwen3 14B 14B	Q4_K	llama.cpp	102.7	4473.1	–	–	16384	–	sourced	–	source
RTX Pro 5000 Blackwell	Qwen3 14B 14B	Q4_K	llama.cpp	98.9	4515.7	–	–	4096	–	sourced	–	source
RTX 4090	Qwen3 14B 14B	Q4_K	llama.cpp	84.4	5265.4	–	–	4096	–	sourced	–	source
RTX Pro 5000 Blackwell	Qwen3 14B 14B	Q4_K	llama.cpp	84.3	3064.2	–	–	16384	–	sourced	–	source
RTX 4090	Qwen3 14B	Q4_K_XL	llama.cpp llama-bench	82.82	–	–	–	4096	2026-08-17	sourced	–	source
RTX 5090	Qwen3 14B 14B	Q4_K	llama.cpp	82.4	2908.4	–	–	32768	–	sourced	–	source
RTX PRO 4500 Blackwell	Qwen3 14B	Q4_K - Medium	llama.cpp (CUDA)	82.32	3861.85	–	–	tg128	2025-11-14	community	–	source
RTX 6000 Ada	Qwen3 14B 14B	Q4_K	llama.cpp	81.2	2912.1	–	–	4096	–	sourced	–	source
RTX 5080	Qwen3 14B 14B	Q4_K	llama.cpp	80.6	3820.5	–	–	4096	–	sourced	–	source
RTX PRO 4500 Blackwell	Qwen3 14B	Q4_K - Medium	llama.cpp (CUDA)	79.37	3583.82	–	–	tg128	2025-11-14	community	–	source
RTX 3090 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	76.2	2817.1	–	–	4096	–	sourced	–	source
RTX 5070 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	74.3	3264.7	–	–	4096	–	sourced	–	source
RTX Pro 5000 Blackwell	Qwen3 14B 14B	Q4_K	llama.cpp	70.8	2069.8	–	–	32768	–	sourced	–	source
RTX 3080 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	69.9	2600.1	–	–	4096	–	sourced	–	source
RTX 4090	Qwen3 14B 14B	Q4_K	llama.cpp	69.8	3452.3	–	–	16384	–	sourced	–	source
V100	Qwen3 14B	Q4_K - Medium	llama.cpp (CUDA)	67.03	1474.29	–	–	tg128	2025-09-22	community	–	source
V100	Qwen3 14B	Q4_K - Medium	llama.cpp (CUDA)	64.87	1536.27	–	–	tg128	2025-09-22	community	–	source
RTX 4080 SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	64.2	3745.0	–	–	4096	–	sourced	–	source
RTX 5080	Qwen3 14B 14B	Q4_K	llama.cpp	64.0	2542.0	–	–	16384	–	sourced	–	source
RTX 4080	Qwen3 14B 14B	Q4_K	llama.cpp	62.0	3574.7	–	–	4096	–	sourced	–	source
RTX 6000 Ada	Qwen3 14B 14B	Q4_K	llama.cpp	58.5	2367.2	–	–	16384	–	sourced	–	source
RTX 4070 Ti SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	58.1	3048.0	–	–	4096	–	sourced	–	source
RTX 5070 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	58.0	2303.9	–	–	16384	–	sourced	–	source
RTX 5090	Qwen3 14B 14B	Q4_K	llama.cpp	57.6	1707.2	–	–	65536	–	sourced	–	source
RTX 3090 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	56.9	1914.8	–	–	16384	–	sourced	–	source
RTX 4090	Qwen3 14B 14B	Q4_K	llama.cpp	55.4	2324.2	–	–	32768	–	sourced	–	source
RTX 5070	Qwen3 14B 14B	Q4_K	llama.cpp	54.2	2144.2	–	–	4096	–	sourced	–	source
RTX A6000	Qwen3 14B 14B	Q4_K	llama.cpp	53.8	1713.3	–	–	4096	–	sourced	–	source
RTX Pro 5000 Blackwell	Qwen3 14B 14B	Q4_K	llama.cpp	53.8	1125.1	–	–	65536	–	sourced	–	source
RTX 4080 SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	52.8	2526.7	–	–	16384	–	sourced	–	source
RTX 3080 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	52.3	1773.9	–	–	16384	–	sourced	–	source
RTX 5080	Qwen3 14B 14B	Q4_K	llama.cpp	51.9	1326.1	–	–	32768	–	sourced	–	source
RTX 4080	Qwen3 14B 14B	Q4_K	llama.cpp	51.2	2295.1	–	–	16384	–	sourced	–	source
RTX 4070 Ti SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	47.2	2003.4	–	–	16384	–	sourced	–	source
RTX 4070 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	45.8	2866.6	–	–	4096	–	sourced	–	source
RTX 4070 SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	45.5	2522.2	–	–	4096	–	sourced	–	source
RTX 5070 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	45.5	1658.2	–	–	32768	–	sourced	–	source
RTX 4080 SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	42.6	1769.3	–	–	32768	–	sourced	–	source
RTX 4070	Qwen3 14B 14B	Q4_K	llama.cpp	42.5	2099.9	–	–	4096	–	sourced	–	source
RTX 6000 Ada	Qwen3 14B 14B	Q4_K	llama.cpp	42.3	1310.2	–	–	32768	–	sourced	–	source
RTX 3090 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	42.1	1232.1	–	–	32768	–	sourced	–	source
M3 Ultra	Qwen3 14B	8bit	oMLX	41.9	616.7	–	–	1024	2026-10-09	community	–	source
RTX 5060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	41.1	1743.0	–	–	4096	–	sourced	–	source
RTX A6000	Qwen3 14B 14B	Q4_K	llama.cpp	40.7	1529.0	–	–	16384	–	sourced	–	source
RTX 4080	Qwen3 14B 14B	Q4_K	llama.cpp	40.6	1395.7	–	–	32768	–	sourced	–	source
RTX 5070	Qwen3 14B 14B	Q4_K	llama.cpp	40.6	1315.2	–	–	16384	–	sourced	–	source
Arc B580	Qwen3 14B	unspecified	llama.cpp (Vulkan)	40.0	–	–	–	–	2026-10-05	community	–	source
Arc B580	Qwen3 14B	unspecified	llama.cpp (Vulkan)	40.0	–	–	–	–	2026-10-05	community	–	source
M3 Ultra	Qwen3 14B	8bit	oMLX	39.9	611.5	–	–	4096	2026-10-09	community	–	source
RTX 4090	Qwen3 14B 14B	Q4_K	llama.cpp	38.8	1398.3	–	–	65536	–	sourced	–	source
RTX 4090	Qwen3 14B	Q4_K_XL	llama.cpp llama-bench	38.73	–	–	–	65536	2026-08-17	sourced	–	source
RTX 4070 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	37.8	1951.4	–	–	16384	–	sourced	–	source
RTX 4070 Ti SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	37.7	1236.9	–	–	32768	–	sourced	–	source
RTX 4070 SUPER	Qwen3 14B 14B	Q4_K	llama.cpp	37.2	1578.1	–	–	16384	–	sourced	–	source
RTX 5090	Qwen3 14B 14B	Q4_K	llama.cpp	37.2	908.4	–	–	131072	–	sourced	–	source
RTX Pro 5000 Blackwell	Qwen3 14B 14B	Q4_K	llama.cpp	36.5	434.2	–	–	131072	–	sourced	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	33.0	941.1	–	–	1024	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	33.0	941.1	–	–	1024	2026-10-14	community	–	source
RTX 5060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	32.9	942.6	–	–	16384	–	sourced	–	source
RTX 4070	Qwen3 14B 14B	Q4_K	llama.cpp	32.7	1355.8	–	–	16384	–	sourced	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	31.5	956.1	–	–	4096	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	31.5	956.1	–	–	4096	2026-10-14	community	–	source
RTX 3060 12GB	Qwen3 14B 14B	Q4_K	llama.cpp	31.2	972.6	–	–	4096	–	sourced	–	source
RTX A6000	Qwen3 14B 14B	Q4_K	llama.cpp	30.8	1063.4	–	–	32768	–	sourced	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	28.6	934.1	–	–	8192	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	28.6	934.1	–	–	8192	2026-10-14	community	–	source
M4 Pro	Qwen3 14B	4bit	oMLX	28.5	248.0	–	–	1024	2026-10-08	community	–	source
RTX 3090 Ti	Qwen3 14B 14B	Q4_K	llama.cpp	28.2	828.5	–	–	65536	–	sourced	–	source
RTX 4060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	27.4	1645.7	–	–	4096	–	sourced	–	source
RTX 6000 Ada	Qwen3 14B 14B	Q4_K	llama.cpp	27.0	727.4	–	–	65536	–	sourced	–	source
M4 Pro	Qwen3 14B	4bit	oMLX	26.6	240.6	–	–	4096	2026-10-08	community	–	source
RTX 5060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	25.9	621.0	–	–	32768	–	sourced	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	25.1	831.7	–	–	16384	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	25.1	831.7	–	–	16384	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	24.7	678.3	–	–	16384	2026-10-13	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	24.7	680.2	–	–	16384	2026-10-13	community	–	source
M4 Pro	Qwen3 14B	4bit	oMLX	24.3	228.0	–	–	8192	2026-10-08	community	–	source
M4 Max	Qwen3 14B	8bit	oMLX	23.5	383.3	–	–	1024	2026-10-12	community	–	source
M4 Max	Qwen3 14B	8bit	oMLX	22.9	371.1	–	–	4096	2026-10-12	community	–	source
RTX 3060 12GB	Qwen3 14B 14B	Q4_K	llama.cpp	22.7	678.2	–	–	16384	–	sourced	–	source
RTX 4060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	22.4	917.6	–	–	16384	–	sourced	–	source
M4 Max	Qwen3 14B	8bit	oMLX	22.0	357.2	–	–	8192	2026-10-12	community	–	source
RTX A6000	Qwen3 14B 14B	Q4_K	llama.cpp	20.5	668.9	–	–	65536	–	sourced	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	20.1	645.5	–	–	32768	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	20.1	645.5	–	–	32768	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	19.6	523.7	–	–	32768	2026-10-13	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	19.6	546.9	–	–	32768	2026-10-13	community	–	source
RTX 4060 Ti 16GB	Qwen3 14B 14B	Q4_K	llama.cpp	17.9	541.4	–	–	32768	–	sourced	–	source
M5 Pro	Qwen3 14B	8bit	oMLX	15.8	641.4	–	–	1024	2026-10-08	community	–	source
RTX 6000 Ada	Qwen3 14B 14B	Q4_K	llama.cpp	15.2	358.6	–	–	131072	–	sourced	–	source
M5 Pro	Qwen3 14B	8bit	oMLX	14.6	691.2	–	–	4096	2026-10-08	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	14.1	502.4	–	–	65536	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	14.1	502.4	–	–	65536	2026-10-14	community	–	source
RTX A6000	Qwen3 14B 14B	Q4_K	llama.cpp	12.1	382.4	–	–	131072	–	sourced	–	source
M2	Qwen3 14B	4bit	oMLX	10.1	70.7	–	–	4096	2026-10-08	community	–	source
M2	Qwen3 14B	4bit	oMLX	9.2	67.1	–	–	8192	2026-10-08	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	9.2	332.1	–	–	131072	2026-10-14	community	–	source
M5 Pro	Qwen3 14B	4bit	oMLX	9.2	332.1	–	–	131072	2026-10-14	community	–	source
M4	Qwen3 14B	4bit	oMLX	4.6	41.6	–	–	4096	2026-10-15	community	–	source
M4	Qwen3 14B	4bit	oMLX	4.2	33.4	–	–	1024	2026-10-15	community	–	source

← all models