Model comparison
jina-embeddings-v5-text-small vs jina-embeddings-v5-text-nano
At Q4_K_M, jina-embeddings-v5-text-nano is the smaller download — 156,966,976 bytes against 396,705,152. At long context the gap widens: jina-embeddings-v5-text-nano's KV cache at 32K is 3.1× smaller, which usually matters more than the difference in weights.
From the file· summed bytes, KV per layer
Side by side
| jina-embeddings-v5-text-small | jina-embeddings-v5-text-nano | |
|---|---|---|
| Parameters | 596M | 212M |
| Architecture | qwen3 | eurobert |
| Layers | 28 | 12 |
| Native context | 32,768 | 8,192 |
| Mixture of experts | no | no |
| Quantizations published | 56 | 42 |
| Smallest quantization | 0.19 GiB | 0.09 GiB |
| Q4_K_M | 0.37 GiB | 0.15 GiB |
| Licence | cc-by-nc-4.0 | cc-by-nc-4.0 |
KV cache by context
the term that decides long-context viability
| Context | jina-embeddings-v5-text-small | jina-embeddings-v5-text-nano | Ratio |
|---|---|---|---|
| 4,096 | 0.44 GiB | 0.14 GiB | 3.11× |
| 8,192 | 0.88 GiB | 0.28 GiB | 3.11× |
| 16,384 | 1.75 GiB | 0.56 GiB | 3.11× |
| 32,768 | 3.50 GiB | 1.13 GiB | 3.11× |
| 65,536 | 7.00 GiB | 2.25 GiB | 3.11× |
| 131,072 | 14.00 GiB | 4.50 GiB | 3.11× |