How much VRAM does meta-llama-3.1-8b-bnb-4bit need?

Q4_K_M is exactly 4,920,734,080 bytes (4.58 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is meta-llama-3.1-8b-bnb-4bit's KV cache?

4.00 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.

unsloth · text

meta-llama-3.1-8b-bnb-4bit

unsloth/meta-llama-3.1-8b-bnb-4bit

meta-llama-3.1-8b-bnb-4bit at Q4_K_M is exactly 4,920,734,080 bytes (4.58 GiB / 4.92 GB) — an effective 4.772 bits per weight, not the nominal 4. Its KV cache at 32K is 4.00 GiB.

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

Parameters

8.2B

Architecture

llama

32 layers

Context

131,072

native (config.json)

License

creativeml-openrail-m

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q3_K_S	3.41 GiB	3,664,499,072	3.554	—	theprint
Q3_K_M	3.74 GiB	4,018,917,760	3.898	—	theprint
Q4_K_S	4.37 GiB	4,692,668,800	4.551	—	theprint
Q4_K_M	4.58 GiB	4,920,734,080	4.772	—	baoqtrinh
Q4_K_M	4.58 GiB	4,920,734,080	4.772	—	theprint
Q5_K_M	5.34 GiB	5,732,987,264	5.560	—	theprint
Q6_K	6.14 GiB	6,596,006,272	6.397	—	theprint
Q8_0	7.95 GiB	8,540,770,688	8.283	—	theprint
F16	14.97 GiB	16,068,891,008	15.584	—	prithivMLmods
F16	14.97 GiB	16,068,891,008	15.584	—	theprint
BF16	14.97 GiB	16,068,891,008	15.584	—	theprint
F16	14.97 GiB	16,068,891,008	15.584	—	baoqtrinh

KV cache by context

computed per layer

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	0.50 GiB	0.50 GiB	—	32 / 0 / 0
8,192	1.00 GiB	1.00 GiB	—	32 / 0 / 0
16,384	2.00 GiB	2.00 GiB	—	32 / 0 / 0
32,768	4.00 GiB	4.00 GiB	—	32 / 0 / 0
65,536	8.00 GiB	8.00 GiB	—	32 / 0 / 0
131,072	16.00 GiB	16.00 GiB	—	32 / 0 / 0

Compare with

same modality, comparable size

vs Qwen3-Coder-30B-A3B-Instruct vs Qwen3.8-27B vs gemma-4-12B-it-qat-q4_0-unquantized vs gemma-4-E4B-it vs Qwen3-30B-A3B-Thinking-2507 vs Qwen3-4B 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 4.32 GiB. The real file is 4.58 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

4096

Vocab

128,256

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does meta-llama-3.1-8b-bnb-4bit need?: Q4_K_M is exactly 4,920,734,080 bytes (4.58 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is meta-llama-3.1-8b-bnb-4bit's KV cache?: 4.00 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 meta-llama-3.1-8b-bnb-4bit 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.