How much VRAM does Dolphin3.0-Llama3.2-1B need?

Q4_K_M is exactly 807,697,440 bytes (0.75 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

How large is Dolphin3.0-Llama3.2-1B's KV cache?

1.00 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.

dphn · text

Dolphin3.0-Llama3.2-1B

dphn/Dolphin3.0-Llama3.2-1B

Dolphin3.0-Llama3.2-1B at Q4_K_M is exactly 807,697,440 bytes (0.75 GiB / 0.81 GB) — an effective 5.229 bits per weight, not the nominal 4. Its KV cache at 32K is 1.00 GiB.

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

Parameters

1.2B

Architecture

llama

16 layers

Context

131,072

native (config.json)

License

llama3.2

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
IQ2_M	0.48 GiB	515,451,392	3.337	—	bartowski
Q2_K	0.54 GiB	580,877,344	3.760	—	bartowski
IQ3_XS	0.58 GiB	621,116,448	4.021	—	bartowski
Q3_K_S	0.60 GiB	641,694,752	4.154	—	bartowski
Q2_K_L	0.60 GiB	644,493,312	4.172	—	bartowski
IQ3_M	0.61 GiB	657,292,320	4.255	—	bartowski
Q3_K_M	0.64 GiB	690,846,752	4.472	—	bartowski
Q3_K_L	0.68 GiB	732,527,648	4.742	—	bartowski
IQ4_XS	0.69 GiB	743,144,480	4.811	—	bartowski
IQ4_NL	0.72 GiB	773,028,896	5.004	—	bartowski
Q4_0	0.72 GiB	773,028,896	5.004	—	bartowski
Q4_K_S	0.72 GiB	775,650,336	5.021	—	bartowski
Q4_K_M	0.75 GiB	807,697,440	5.229	—	bartowski
Q4_1	0.77 GiB	831,749,152	5.384	—	bartowski
Q4_K_L	0.81 GiB	871,313,408	5.640	—	bartowski
Q5_K_S	0.83 GiB	892,566,560	5.778	—	bartowski
Q5_K_M	0.85 GiB	911,506,464	5.901	—	bartowski
Q5_K_L	0.91 GiB	975,122,432	6.312	—	bartowski
Q6_K	0.95 GiB	1,021,803,552	6.615	—	bartowski
Q6_K_L	1.01 GiB	1,085,419,520	7.026	—	bartowski
Q8_0	1.23 GiB	1,321,086,976	8.552	—	bartowski
F16	2.31 GiB	2,479,603,456	16.052	—	bartowski
F32	4.61 GiB	4,951,104,992	32.051	—	bartowski

KV cache by context

computed per layer

Context	KV cache (f16)●	Flat formula	Overstated by	Full / windowed / recurrent
4,096	0.13 GiB	0.13 GiB	—	16 / 0 / 0
8,192	0.25 GiB	0.25 GiB	—	16 / 0 / 0
16,384	0.50 GiB	0.50 GiB	—	16 / 0 / 0
32,768	1.00 GiB	1.00 GiB	—	16 / 0 / 0
65,536	2.00 GiB	2.00 GiB	—	16 / 0 / 0
131,072	4.00 GiB	4.00 GiB	—	16 / 0 / 0

Compare with

same modality, comparable size

vs Qwen3-4B vs HyperCLOVAX-SEED-Text-Instruct-1.5B vs Llama-3.2-1B-Instruct vs ced-base vs gemma-3-1b-it vs Qwen3-1.7B 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 0.65 GiB. The real file is 0.75 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

Hidden size

2048

Vocab

128,258

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does Dolphin3.0-Llama3.2-1B need?: Q4_K_M is exactly 807,697,440 bytes (0.75 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is Dolphin3.0-Llama3.2-1B's KV cache?: 1.00 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 Dolphin3.0-Llama3.2-1B 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.