Model comparison

whisper-base.en vs whisper-small

At Q4_K_M, whisper-base.en is the smaller download — 58,794,272 bytes against 171,630,656.

From the file· summed bytes, KV per layer

Side by side

whisper-base.enwhisper-small
Parameters73M242M
Architecturewhisperwhisper
Layers612
Native context
Mixture of expertsnono
Quantizations published619
Smallest quantization0.05 GiB0.08 GiB
Q4_K_M0.05 GiB0.16 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextwhisper-base.enwhisper-smallRatio
4,096
8,192
16,384
32,768
65,536
131,072