Model comparison

jina-reranker-v1-tiny-en vs nomic-embed-text-v1.5

At Q4_K_M, jina-reranker-v1-tiny-en is the smaller download — 33,257,536 bytes against 84,106,624. At long context the gap widens: jina-reranker-v1-tiny-en's KV cache at 32K is 6.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

jina-reranker-v1-tiny-ennomic-embed-text-v1.5
Parameters33M137M
Architecturejina-bert-v2nomic-bert
Layers412
Native context8,1922,048
Mixture of expertsnono
Quantizations published1239
Smallest quantization0.03 GiB0.05 GiB
Q4_K_M0.03 GiB0.08 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextjina-reranker-v1-tiny-ennomic-embed-text-v1.5Ratio
4,0960.02 GiB0.14 GiB6.00×
8,1920.05 GiB0.28 GiB6.00×
16,3840.09 GiB0.56 GiB6.00×
32,7680.19 GiB1.13 GiB6.00×
65,5360.38 GiB2.25 GiB6.00×
131,0720.75 GiB4.50 GiB6.00×