How much VRAM does whisper-tiny need?

Q4_K_M is exactly 43,621,792 bytes (0.04 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

openai · audio asr

whisper-tiny

openai/whisper-tiny

whisper-tiny at Q4_K_M is exactly 43,621,792 bytes (0.04 GiB / 0.04 GB) — an effective 9.242 bits per weight, not the nominal 4.

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

Parameters

38M

Architecture

whisper

4 layers

Context

—

native (config.json)

License

apache-2.0

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q2_K	0.02 GiB	16,783,264	3.556	—	oxide-lab
Q3_K	0.02 GiB	20,483,311	4.340	—	oxide-lab
Q4_0	0.02 GiB	22,087,104	4.679	—	oxide-lab
Q4_1	0.02 GiB	24,414,464	5.173	—	oxide-lab
Q4_0	0.02 GiB	25,321,834	5.365	—	oxide-lab
Q4_K	0.02 GiB	25,321,834	5.365	—	oxide-lab
Q4_1	0.03 GiB	27,598,786	5.847	—	oxide-lab
Q5_K	0.03 GiB	29,875,738	6.330	—	oxide-lab
Q5_0	0.03 GiB	29,875,738	6.330	—	oxide-lab
Q5_1	0.03 GiB	32,152,690	6.812	—	oxide-lab
Q6_K	0.03 GiB	34,714,261	7.355	—	oxide-lab
Q8_0	0.04 GiB	43,537,450	9.224	—	oxide-lab
Q4_K_M	0.04 GiB	43,621,792	9.242	—	handy-computer
Q5_K_M	0.04 GiB	44,211,616	9.367	—	handy-computer
Q6_K	0.04 GiB	44,838,304	9.499	—	handy-computer
Q8_0	0.04 GiB	45,981,088	9.742	—	handy-computer
F16	0.07 GiB	80,135,360	16.977	—	handy-computer
F32	0.14 GiB	152,997,824	32.414	—	handy-computer

Compare with

same modality, comparable size

vs whisper-base vs moonshine-streaming-tiny 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 0.02 GiB. The real file is 0.04 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

—

Vocab

51,865

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does whisper-tiny need?: Q4_K_M is exactly 43,621,792 bytes (0.04 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of whisper-tiny 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.