ModelCards

jina-embeddings-v5-text-small

Embedding modelOpen weightsCard summarized

SOTA multilingual embeddings with task-specific adapters

Source: jina-embeddings-v5-text: Task-Targeted Embedding Distillation (technical report, arXiv 2602.15547)

Use it via API

$0.05 / 1M tok
Code exampleCompare

Where to call it

  • Jina Embeddings APIjina-embeddings-v5-text-small
  • Self-host (weights)jinaai/jina-embeddings-v5-text-small
  • Google Vertex AIjina-embeddings-v5-text-small

Supports

  • Batch API
Full feature support

Good for

  • Multilingual text retrieval and RAG over long documents (32K tokens)
  • Semantic similarity, clustering and classification via task adapters
  • Storage-constrained indexes using truncated or binary vectors

Watch out for

Overview

Jina AI · 01Text, front

v5 Text Small

SOTA multilingual embeddings with task-specific adapters

Typical price
$0.05 / 1M tok
Text input, Jina API (1B-token pack)
Dimensions
1024
Max input
32,768 tokens
Modes
text embeddings, retrieval (query/passage)
Released
18 Feb 2026

jina-embeddings-v5-text-small is a 677M-parameter multilingual text embedding model built on Qwen3-0.6B-Base. It is trained by distilling Qwen3-Embedding-4B and then adding four task-specific LoRA adapters (retrieval, text matching, clustering, classification). It takes up to 32K tokens, outputs 1024-dimension vectors that can be truncated, and is designed to stay accurate under binary quantization. tech report abstract

jina-embeddings-v5-text-small
Typical price
$0.05 / 1M tok
Dimensions
1024 (Matryoshka: 32, 64, 128, 256, 512, 768, 1024; API accepts 1–1024)
Max input
32,768 tokens
Languages
93 languages supported (trained on 32)
Weights
Open · CC BY-NC 4.0 (commercial self-hosting needs a Jina On-Prem license from Elastic)
Released
18 Feb 2026
  • text embeddings
  • retrieval (query/passage)
  • text matching
  • clustering
  • classification