model comparison

vs

jina-embeddings-v5-text-nano vs jina-embeddings-v5-text-small

compared on price, indexing cost, vector size and input limits.

the short answer

  • on RTEB, as each maker reports it, jina-embeddings-v5-text-nano scores 64.08 and jina-embeddings-v5-text-small 66.84.
  • Jina v5 Text Nano is cheaper: $0.02/M against $0.05/M.
  • Jina v5 Text Small takes longer inputs (32k tokens against 8k tokens), so it needs less chunking.
  • Jina v5 Text Nano returns smaller vectors (768 against 1024 dimensions), which cuts index storage by 25%.

side-by-side specs

specjina-embeddings-v5-text-nanojina-embeddings-v5-text-small
providerJina AIJina AI
price$0.02/M$0.05/M
max input8k tokens32k tokens
default dimensions7681,024
storage per 1M vectors3.1 GB4.1 GB
acceptstexttext
multilingualyesyes
weightsopen (CC BY-NC 4.0)open (CC BY-NC 4.0)
RTEB (maker-reported)64.0866.84
BEIR (maker-reported)56.0656.67
LongEmbed (maker-reported)63.6566.39

same work, both models

workloadjina-embeddings-v5-text-nanojina-embeddings-v5-text-small
index 10,000 help articles~800 tokens each, 8M tokensJina v5 Text Small costs 2.5× as much$0.16$0.40
index a 1M-line codebase~10 tokens a line, 10M tokens$0.20$0.50
embed 1M search queries~20 tokens each, 20M tokens$0.40$1.00

at listed rates. re-embedding an index costs the same again.

when to pick which

jina-embeddings-v5-text-nano

Jina v5 Text Nano is the smaller of Jina AI's two v5 text embedding models, a 239M-parameter multilingual model built on EuroBERT-210M. it costs less than jina-embeddings-v5-text-small and reads 8K tokens against Small's 32K.

pick it if

  • you want the cheapest Jina v5 text embeddings for multilingual search
  • Jina's reported English MTEB 71.0, near v5-text-small's 71.7, is enough
  • you want 768-dimension vectors that truncate down to 32 dimensions

skip it if

  • your documents exceed 8K tokens; v5-text-small reads 32K
  • you embed images, audio or video; this model reads text only

jina-embeddings-v5-text-small

Jina v5 Text Small is Jina AI's 677M-parameter multilingual text embedding model, built on Qwen3-0.6B-Base and distilled from Qwen3-Embedding-4B. it is the larger of the two v5 text models, above v5-text-nano; v5-omni-small is the multimodal counterpart.

pick it if

  • you embed long documents, up to 32K tokens per input
  • you need multilingual retrieval with vectors truncatable from 1024 down to 32
  • you want open weights to self-host for non-commercial work

skip it if

  • you need image input; v5-omni-small covers multimodal at the same price
  • short inputs suffice and v5-text-nano costs less per token

faq

which is cheaper, Jina v5 Text Nano or Jina v5 Text Small?

Jina v5 Text Nano is cheaper: $0.02/M against $0.05/M.

which handles longer documents?

Jina v5 Text Small takes longer inputs (32k tokens against 8k tokens), so it needs less chunking.

can i switch from Jina v5 Text Nano to Jina v5 Text Small without re-indexing?

no. vectors from different embedding models are not comparable, so a switch means re-embedding the whole corpus.

try either model

switch between models inside the same octomind session.

prices from the catalog of 2026-11-18