Model comparison

nomic-embed-text-v1 vs nomic-embed-text-v1.5

At Q4_K_M, nomic-embed-text-v1.5 is the smaller download — 84,106,624 bytes against 84,106,624.

From the file· summed bytes, KV per layer

Side by side

nomic-embed-text-v1nomic-embed-text-v1.5
Parameters137M137M
Architecturenomic-bertnomic-bert
Layers1212
Native context8,1922,048
Mixture of expertsnono
Quantizations published1439
Smallest quantization0.05 GiB0.05 GiB
Q4_K_M0.08 GiB0.08 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextnomic-embed-text-v1nomic-embed-text-v1.5Ratio
4,0960.14 GiB0.14 GiB1.00×
8,1920.28 GiB0.28 GiB1.00×
16,3840.56 GiB0.56 GiB1.00×
32,7681.13 GiB1.13 GiB1.00×
65,5362.25 GiB2.25 GiB1.00×
131,0724.50 GiB4.50 GiB1.00×