Model comparison

parakeet-unified-en-0.6b vs parakeet-tdt-0.6b-v2

At Q4_K_M, parakeet-tdt-0.6b-v2 is the smaller download — 475,491,840 bytes against 477,274,496.

From the file· summed bytes, KV per layer

Side by side

parakeet-unified-en-0.6bparakeet-tdt-0.6b-v2
Parameters618M618M
Architectureparakeetparakeet
Layers
Native context
Mixture of expertsnono
Quantizations published68
Smallest quantization0.44 GiB0.37 GiB
Q4_K_M0.44 GiB0.44 GiB
Licencecc-by-4.0cc-by-4.0

KV cache by context

the term that decides long-context viability
Contextparakeet-unified-en-0.6bparakeet-tdt-0.6b-v2Ratio
4,096
8,192
16,384
32,768
65,536
131,072