How much VRAM does Qwen3 8B need?

At Q4_K_M the weights are 4.6 GB; with an 8K context KV cache and runtime overhead it needs 6.3 GB, so a 10 GB card runs it with headroom.

Can a GeForce RTX 4070 run Qwen3 8B?

Yes — Q6_K fits with headroom at 8K context, about 49 tokens per second estimated.

Can a GeForce RTX 4090 run Qwen3 8B?

Yes — Q8_0 fits with headroom at 8K context, about 75 tokens per second estimated.

Can a GeForce RTX 3060 12 GB run Qwen3 8B?

Yes — Q6_K fits with headroom at 8K context, about 35 tokens per second estimated.

Can a M3 Max · 64 GB run Qwen3 8B?

Yes — Q8_0 fits with headroom at 8K context, about 28 tokens per second estimated.

Hardware requirements

What hardware do I need to run Qwen3 8B?

8.19B 128K context Apache 2.0

At Q4_K_M the weights are 4.6 GB; with an 8K KV cache and runtime overhead that is 6.3 GB, so it wants a 10 GB card to run with headroom — in practice 10 GB VRAM: GeForce RTX 3080 and 1 other. Every figure below is computed from the model’s published shape against each device’s usable memory; tokens per second are estimates. The method.

Minimum practical 8 GB VRAM

GeForce RTX 3070 Ti — Q4_K_M, 6.3 GB of 7.0 GB, ~80 tok/s, up to 12K context.

Also: Arc A750, GeForce RTX 5060, GeForce RTX 3070, GeForce RTX 3060 Ti, GeForce RTX 4060 Ti +2 more

the smallest card that holds it at all, at Q3_K_M — the lowest quantisation worth running

Recommended 10 GB VRAM

GeForce RTX 3080 — Q4_K_M, 6.3 GB of 8.8 GB, ~100 tok/s, up to 25K context.

Also: Arc B570

runs the recommended quantisation with headroom at 8K context

High quality 16 GB VRAM

GeForce RTX 5080 — Q8_0, 9.8 GB of 14.4 GB, ~72 tok/s, up to 40K context.

Also: GeForce RTX 5070 Ti, GeForce RTX 4080 Super, GeForce RTX 4080, GeForce RTX 4070 Ti Super, Radeon RX 9070 XT +12 more

near-lossless Q8_0 with headroom

Long context 32 GB VRAM

GeForce RTX 5090 — Q4_K_M, 23.2 GB of 30.4 GB, ~236 tok/s at 128K.

Also: RTX 5000 Ada, Radeon Pro W7800

128K tokens (or the model’s maximum) at the recommended quantisation

Apple Silicon 16 GB unified

M1 · 16 GB — Q4_K_M, 6.3 GB of 10.7 GB, ~8.3 tok/s, up to 39K context.

smallest Mac that runs the recommended quantisation with headroom

CPU only 32 GB RAM

CPU only · DDR4 dual-channel — Q8_0, 9.8 GB of 25.6 GB, ~2.6 tok/s, up to 120K context.

no GPU — weights stream from system RAM

How much VRAM at each quantisation

Quant	Weights	+KV 8K	+KV 32K	+KV 128K	Total @ 8K	Needs	Quality
F16	15.3 GB	1.13 GB	4.50 GB	18.0 GB	17.0 GB	24 GB card	Reference
Q8_0	8.1 GB	1.13 GB	4.50 GB	18.0 GB	9.8 GB	16 GB card	−0.1% ppl
Q6_K	6.3 GB	1.13 GB	4.50 GB	18.0 GB	8.0 GB	11 GB card	−0.4% ppl
Q5_K_M	5.4 GB	1.13 GB	4.50 GB	18.0 GB	7.1 GB	10 GB card	−0.8% ppl
Q4_K_M	4.6 GB	1.13 GB	4.50 GB	18.0 GB	6.3 GB	10 GB card	−1.9% ppl
Q3_K_M	3.7 GB	1.13 GB	4.50 GB	18.0 GB	5.5 GB	8 GB card	−5.4% ppl
Q2_K	3.2 GB	1.13 GB	4.50 GB	18.0 GB	4.9 GB	8 GB card	−15% ppl

Totals add 0.6 GB runtime overhead and an f16 KV cache; "needs" is the smallest discrete card class whose usable memory (after display and driver reserve) leaves 15% headroom. A q8_0 cache roughly halves the KV columns.

Every device, checked

System RAM for offload:

16 32 64 128 256

NVIDIA · GeForce

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
GeForce RTX 5090	32 GB	Q8_0	Q8_0	128K	236	yes	Check
GeForce RTX 5080	16 GB	Q8_0	Q8_0	65K	126	no	Check
GeForce RTX 5070 Ti	16 GB	Q8_0	Q8_0	65K	118	no	Check
GeForce RTX 5070	12 GB	Q6_K	Q8_0	38K	88	no	Check
GeForce RTX 5060 Ti 16 GB	16 GB	Q8_0	Q8_0	65K	59	no	Check
GeForce RTX 5060	8 GB	Q3_K_M	Q4_K_M	12K	59	no	Check
GeForce RTX 4090	24 GB	Q8_0	Q8_0	122K	133	no	Check
GeForce RTX 4080 Super	16 GB	Q8_0	Q8_0	65K	97	no	Check
GeForce RTX 4080	16 GB	Q8_0	Q8_0	65K	94	no	Check
GeForce RTX 4070 Ti Super	16 GB	Q8_0	Q8_0	65K	88	no	Check
GeForce RTX 4070 Ti	12 GB	Q6_K	Q8_0	38K	66	no	Check
GeForce RTX 4070 Super	12 GB	Q6_K	Q8_0	38K	66	no	Check
GeForce RTX 4070	12 GB	Q6_K	Q8_0	38K	66	no	Check
GeForce RTX 4060 Ti 16 GB	16 GB	Q8_0	Q8_0	65K	38	no	Check
GeForce RTX 4060 Ti	8 GB	Q3_K_M	Q4_K_M	12K	38	no	Check
GeForce RTX 4060	8 GB	Q3_K_M	Q4_K_M	12K	36	no	Check
GeForce RTX 3090 Ti	24 GB	Q8_0	Q8_0	122K	133	no	Check
GeForce RTX 3090	24 GB	Q8_0	Q8_0	122K	123	no	Check
GeForce RTX 3080 Ti	12 GB	Q6_K	Q8_0	38K	120	no	Check
GeForce RTX 3080 12 GB	12 GB	Q6_K	Q8_0	38K	120	no	Check
GeForce RTX 3080	10 GB	Q5_K_M	Q6_K	25K	100	no	Check
GeForce RTX 3070 Ti	8 GB	Q3_K_M	Q4_K_M	12K	80	no	Check
GeForce RTX 3070	8 GB	Q3_K_M	Q4_K_M	12K	59	no	Check
GeForce RTX 3060 Ti	8 GB	Q3_K_M	Q4_K_M	12K	59	no	Check
GeForce RTX 3060 12 GB	12 GB	Q6_K	Q8_0	38K	47	no	Check
GeForce RTX 2080 Ti	11 GB	Q6_K	Q6_K	31K	81	no	Check
GeForce RTX 2060 12 GB	12 GB	Q6_K	Q8_0	38K	44	no	Check
GeForce GTX 1080 Ti	11 GB	Q6_K	Q6_K	31K	64	no	Check

NVIDIA · GeForce laptop

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
GeForce RTX 5090 Laptop	24 GB	Q8_0	Q8_0	122K	118	no	Check
GeForce RTX 4090 Laptop	16 GB	Q8_0	Q8_0	65K	76	no	Check
GeForce RTX 4080 Laptop	12 GB	Q6_K	Q8_0	38K	57	no	Check
GeForce RTX 4070 Laptop	8 GB	Q3_K_M	Q4_K_M	12K	34	no	Check

NVIDIA · Workstation

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
RTX 6000 Ada	48 GB	Q8_0	Q8_0	128K	126	yes	Check
RTX 5000 Ada	32 GB	Q8_0	Q8_0	128K	76	yes	Check
RTX 4000 Ada	20 GB	Q8_0	Q8_0	93K	47	no	Check
RTX 2000 Ada	16 GB	Q8_0	Q8_0	65K	29	no	Check
RTX A6000	48 GB	Q8_0	Q8_0	128K	101	yes	Check
RTX A5000	24 GB	Q8_0	Q8_0	122K	101	no	Check
RTX A4000	16 GB	Q8_0	Q8_0	65K	59	no	Check

NVIDIA · Data centre

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
L40S	48 GB	Q8_0	Q8_0	128K	114	yes	Check
L4	24 GB	Q8_0	Q8_0	122K	39	no	Check
A10	24 GB	Q8_0	Q8_0	122K	79	no	Check
A100 40 GB	40 GB	Q8_0	Q8_0	128K	204	yes	Check
A100 80 GB	80 GB	Q8_0	Q8_0	128K	268	yes	Check
H100 PCIe	80 GB	Q8_0	Q8_0	128K	263	yes	Check
H100 SXM	80 GB	Q8_0	Q8_0	128K	440	yes	Check
H200 SXM	141 GB	Q8_0	Q8_0	128K	631	yes	Check
Tesla P40	24 GB	Q8_0	Q8_0	122K	45	no	Check

AMD · Radeon

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Radeon RX 9070 XT	16 GB	Q8_0	Q8_0	65K	85	no	Check
Radeon RX 9070	16 GB	Q8_0	Q8_0	65K	85	no	Check
Radeon RX 7900 XTX	24 GB	Q8_0	Q8_0	122K	126	no	Check
Radeon RX 7900 XT	20 GB	Q8_0	Q8_0	93K	105	no	Check
Radeon RX 7900 GRE	16 GB	Q8_0	Q8_0	65K	76	no	Check
Radeon RX 7800 XT	16 GB	Q8_0	Q8_0	65K	82	no	Check
Radeon RX 7700 XT	12 GB	Q6_K	Q8_0	38K	57	no	Check
Radeon RX 7600 XT	16 GB	Q8_0	Q8_0	65K	38	no	Check
Radeon RX 6900 XT	16 GB	Q8_0	Q8_0	65K	67	no	Check
Radeon RX 6800 XT	16 GB	Q8_0	Q8_0	65K	67	no	Check
Radeon RX 6700 XT	12 GB	Q6_K	Q8_0	38K	50	no	Check

AMD · Radeon Pro

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Radeon Pro W7900	48 GB	Q8_0	Q8_0	128K	114	yes	Check
Radeon Pro W7800	32 GB	Q8_0	Q8_0	128K	76	yes	Check

AMD · Instinct

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Instinct MI210	64 GB	Q8_0	Q8_0	128K	215	yes	Check
Instinct MI300X	192 GB	Q8_0	Q8_0	128K	697	yes	Check

AMD · Ryzen AI Max

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Ryzen AI Max+ 395 · 128 GB	128 GB	Q8_0	Q8_0	128K	31	yes	Check
Ryzen AI Max+ 395 · 64 GB	64 GB	Q8_0	Q8_0	128K	31	yes	Check

Apple · Apple Silicon

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
M1 · 16 GB	16 GB	Q6_K	Q8_0	39K	8.3	no	Check
M1 Pro · 32 GB	32 GB	Q8_0	Q8_0	128K	24	tight	Check
M1 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	49	yes	Check
M1 Ultra · 128 GB	128 GB	Q8_0	Q8_0	128K	97	yes	Check
M2 · 24 GB	24 GB	Q8_0	Q8_0	90K	12	no	Check
M2 Pro · 32 GB	32 GB	Q8_0	Q8_0	128K	24	tight	Check
M2 Max · 96 GB	96 GB	Q8_0	Q8_0	128K	49	yes	Check
M2 Ultra · 192 GB	192 GB	Q8_0	Q8_0	128K	97	yes	Check
M3 · 24 GB	24 GB	Q8_0	Q8_0	90K	12	no	Check
M3 Pro · 36 GB	36 GB	Q8_0	Q8_0	128K	18	tight	Check
M3 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	49	yes	Check
M3 Max · 128 GB	128 GB	Q8_0	Q8_0	128K	49	yes	Check
M3 Ultra · 256 GB	256 GB	Q8_0	Q8_0	128K	97	yes	Check
M3 Ultra · 512 GB	512 GB	Q8_0	Q8_0	128K	97	yes	Check
M4 · 32 GB	32 GB	Q8_0	Q8_0	128K	15	tight	Check
M4 Pro · 48 GB	48 GB	Q8_0	Q8_0	128K	33	yes	Check
M4 Pro · 64 GB	64 GB	Q8_0	Q8_0	128K	33	yes	Check
M4 Max · 64 GB	64 GB	Q8_0	Q8_0	128K	50	yes	Check
M4 Max · 128 GB	128 GB	Q8_0	Q8_0	128K	66	yes	Check

Intel · Arc

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
Arc B580	12 GB	Q6_K	Q8_0	38K	60	no	Check
Arc B570	10 GB	Q5_K_M	Q6_K	25K	50	no	Check
Arc A770 16 GB	16 GB	Q8_0	Q8_0	65K	74	no	Check
Arc A750	8 GB	Q3_K_M	Q4_K_M	12K	67	no	Check

CPU · CPU + system RAM

Device	Memory	Best quant with headroom	Fits at all	Max context @ Q4_K_M	Tok/s est.	128K
CPU only · DDR4 dual-channel	32 GB	Q8_0	Q8_0	128K	4.6	tight	Check
CPU only · DDR5 dual-channel	64 GB	Q8_0	Q8_0	128K	8.2	yes	Check
CPU only · DDR5 8-channel server	256 GB	Q8_0	Q8_0	128K	28	yes	Check