jina-embeddings-v5-text-nano-retrieval-mlx

MLX port of jina-embeddings-v5-text-nano-retrieval for efficient inference on Apple Silicon. For the full-size model, see jina-embeddings-v5-text-small-retrieval-mlx.

Elastic Inference Service | ArXiv | Blog

Model Overview

Feature	Value
Parameters	239M
Supported Task	retrieval
Max Sequence Length	8192
Embedding Dimension	768
Pooling Strategy	Last-token pooling
Base Model	EuroBERT/EuroBERT-210m
Format	MLX float16 safetensors

For training details and evaluation results, see our technical report.

Requirements

pip install mlx tokenizers

Usage

git clone https://huggingface.co/jinaai/jina-embeddings-v5-text-nano-retrieval-mlx
cd jina-embeddings-v5-text-nano-retrieval-mlx

import mlx.core as mx
from tokenizers import Tokenizer
from model import JinaEmbeddingModel
import json

with open("config.json") as f:
    config = json.load(f)

model = JinaEmbeddingModel(config)
weights = mx.load("model.safetensors")
model.load_weights(list(weights.items()))

tokenizer = Tokenizer.from_file("tokenizer.json")

# Encode query
query_embeddings = model.encode(
    ["Overview of climate change impacts on coastal cities"],
    tokenizer,
    task_type="retrieval.query",
)

# Encode document
document_embeddings = model.encode(
    ["Climate change has led to rising sea levels, increased frequency of extreme weather events..."],
    tokenizer,
    task_type="retrieval.passage",
)

License

jina-embeddings-v5-text-nano is licensed under CC BY-NC 4.0. For commercial use, please contact sales@jina.ai.

Citation

If you find jina-embeddings-v5-text-nano useful in your research, please cite the following paper:

@article{akram2026jina,
  title={jina-embeddings-v5-text: Task-Targeted Embedding Distillation},
  author={Mohammad Kalim Akram and Saba Sturua and Nastia Havriushenko and Quentin Herreros and Michael G{\"u}nther and Maximilian Werk and Han Xiao},
  journal={arXiv preprint arXiv:2602.15547},
  year={2026}
}