Model comparison

whisper-medium vs whisper-large-v3

At Q4_K_M, whisper-medium is the smaller download — 504,102,848 bytes against 997,303,008.

From the file· summed bytes, KV per layer

Side by side

whisper-mediumwhisper-large-v3
Parameters764M1.5B
Architecturewhisperwhisper
Layers2432
Native context
Mixture of expertsnono
Quantizations published1918
Smallest quantization0.25 GiB0.49 GiB
Q4_K_M0.47 GiB0.93 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextwhisper-mediumwhisper-large-v3Ratio
4,096
8,192
16,384
32,768
65,536
131,072