Jina Embedder

Generate Jina retrieval embeddings with explicit passage and query tasks.

JinaEmbedder defaults to jina-embeddings-v3 with 1024 dimensions. Pass Jina API fields through request_params.

jina_embedder.py

from agno.knowledge.embedder.jina import JinaEmbedder

passage_embedder = JinaEmbedder(
    request_params={"task": "retrieval.passage"},
)
query_embedder = JinaEmbedder(
    request_params={"task": "retrieval.query"},
)

passage_vector = passage_embedder.get_embedding(
    "The quick brown fox jumps over the lazy dog."
)
query_vector = query_embedder.get_embedding("Which animal jumps?")

print(f"Passage dimensions: {len(passage_vector)}")
print(f"Query dimensions: {len(query_vector)}")

JinaEmbedder applies one request_params dictionary to every call. A single instance used by a vector database therefore applies the same task to document insertion and query search. Jina’s asymmetric retrieval tasks distinguish retrieval.passage from retrieval.query.

Run the Example

1. Set up your virtual environment

   Mac

   uv venv --python 3.12
   source .venv/bin/activate
   

   uv venv --python 3.12
   .venv\Scripts\activate
   
2. Export the API key

   export JINA_API_KEY=your_jina_api_key_here
   
3. Install dependencies

   uv pip install -U agno aiohttp requests
   
4. Run the example

   python jina_embedder.py
   

Developer Resources

• Embedders overview
• Jina Embeddings API
• Jina Embeddings v3 tasks