Model comparison

granite-4.0-1b-speech vs Voxtral-Mini-4B-Realtime-2602

At Q4_K_M, granite-4.0-1b-speech is the smaller download — 1,602,904,800 bytes against 2,830,493,984.

From the file· summed bytes, KV per layer

Side by side

granite-4.0-1b-speechVoxtral-Mini-4B-Realtime-2602
Parameters2.3B4.4B
Architecturegranite_speechvoxtral_realtime
Layers4026
Native context4,096131,072
Mixture of expertsnono
Quantizations published68
Smallest quantization1.49 GiB2.35 GiB
Q4_K_M1.49 GiB2.64 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextgranite-4.0-1b-speechVoxtral-Mini-4B-Realtime-2602Ratio
4,0960.31 GiB
8,1920.63 GiB
16,3841.25 GiB
32,7682.50 GiB
65,5365.00 GiB
131,07210.00 GiB