How much VRAM does Qwen3-14B need?

Q4_K_M is exactly 9,001,752,960 bytes (8.38 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is Qwen3-14B's KV cache?

5.00 GiB at 32K context with an f16 cache, computed per layer from the model's architecture. Quantizing the cache to q8_0 roughly halves it.

Qwen · text

Qwen3-14B

Qwen/Qwen3-14B

Qwen3-14B at Q4_K_M is exactly 9,001,752,960 bytes (8.38 GiB / 9.00 GB) — an effective 4.876 bits per weight, not the nominal 4. Its KV cache at 32K is 5.00 GiB.

From the file· summed from 1 file(s)From the file· KV per layer

Parameters

14.8B

Architecture

qwen3

40 layers

Context

40,960

native (config.json)

License

—

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
UD-IQ1_S	3.56 GiB	3,827,369,344	2.073	—	unsloth
UD-IQ1_M	3.79 GiB	4,064,814,464	2.202	—	unsloth
UD-IQ2_XXS	4.16 GiB	4,470,113,664	2.421	—	unsloth
UD-IQ2_M	5.05 GiB	5,425,956,224	2.939	—	unsloth
Q2_K	5.36 GiB	5,753,984,384	3.117	—	unsloth
Q2_K_L	5.53 GiB	5,936,307,584	3.216	—	unsloth
UD-IQ3_XXS	5.60 GiB	6,013,732,224	3.258	—	unsloth
Q3_K_S	6.20 GiB	6,657,106,304	3.606	—	unsloth
Q3_K_M	6.82 GiB	7,321,313,664	3.966	443	unsloth
Q3_K_L	7.36 GiB	7,900,651,296	4.280	—	lmstudio-community
IQ4_XS	7.58 GiB	8,135,040,384	4.407	443	unsloth
IQ4_NL	7.95 GiB	8,541,363,584	4.627	—	unsloth
Q4_0	7.96 GiB	8,543,001,984	4.628	443	unsloth
Q4_K_S	7.98 GiB	8,573,476,224	4.644	—	unsloth
Q4_K_M	8.38 GiB	9,001,752,960	4.876	443	Qwen
Q4_K_M	8.38 GiB	9,001,753,376	4.876	—	lmstudio-community
Q4_K_M	8.38 GiB	9,001,753,984	4.876	—	unsloth
Q4_1	8.74 GiB	9,389,522,304	5.086	—	unsloth
Q5_0	9.56 GiB	10,263,894,400	5.560	—	Qwen
Q5_K_S	9.56 GiB	10,263,895,424	5.560	—	unsloth
Q5_K_M	9.79 GiB	10,514,569,568	5.696	443	Qwen
Q5_K_M	9.79 GiB	10,514,570,624	5.696	—	unsloth
Q6_K	11.29 GiB	12,121,937,248	6.566	443	Qwen
Q6_K	11.29 GiB	12,121,937,696	6.566	—	lmstudio-community
Q6_K	11.29 GiB	12,121,938,304	6.566	—	unsloth
Q8_0	14.62 GiB	15,698,533,728	8.504	443	Qwen
Q8_0	14.62 GiB	15,698,534,176	8.504	—	lmstudio-community
Q8_0	14.62 GiB	15,698,534,784	8.504	—	unsloth
BF16	27.51 GiB	29,543,424,160	16.004	—	unsloth

KV cache by context

computed per layer

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	0.63 GiB	0.63 GiB	—	40 / 0 / 0
8,192	1.25 GiB	1.25 GiB	—	40 / 0 / 0
16,384	2.50 GiB	2.50 GiB	—	40 / 0 / 0
32,768	5.00 GiB	5.00 GiB	—	40 / 0 / 0
65,536	10.00 GiB	10.00 GiB	—	40 / 0 / 0
131,072	20.00 GiB	20.00 GiB	—	40 / 0 / 0

Compare with

same modality, comparable size

vs Qwen3-Coder-30B-A3B-Instruct vs Qwen3.6-27B vs Qwen3.8-27B vs gemma-4-12B-it-qat-q4_0-unquantized vs gemma-4-E4B-it vs Qwen3-30B-A3B-Thinking-2507 All alternatives →Will it fit on my card? →

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 7.74 GiB. The real file is 8.38 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json

Layers

Attention heads

KV heads

Head dim

128

Hidden size

5120

Vocab

151,936

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

false

Questions people ask

How much VRAM does Qwen3-14B need?: Q4_K_M is exactly 9,001,752,960 bytes (8.38 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Qwen3-14B's KV cache?: 5.00 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Which quantization of Qwen3-14B should I use?: Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.