> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agno.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini Embedder

> Generate Gemini embeddings with an explicit retrieval task type and vector dimension.

`GeminiEmbedder` defaults to `gemini-embedding-001`, 1536 dimensions, and the `RETRIEVAL_QUERY` task type.

```python gemini_embedder.py theme={null}
from math import sqrt

from agno.knowledge.embedder.google import GeminiEmbedder


def normalize(vector: list[float]) -> list[float]:
    magnitude = sqrt(sum(value * value for value in vector))
    return [value / magnitude for value in vector] if magnitude else vector


document_embedder = GeminiEmbedder(
    id="gemini-embedding-001",
    dimensions=1536,
    task_type="RETRIEVAL_DOCUMENT",
)
query_embedder = GeminiEmbedder(
    id="gemini-embedding-001",
    dimensions=1536,
    task_type="RETRIEVAL_QUERY",
)

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

print(f"Document dimensions: {len(document_vector)}")
print(f"Query dimensions: {len(query_vector)}")
```

<Warning>
  `GeminiEmbedder` currently applies one configured `task_type` to every call. A single instance used by a vector database therefore applies the same task type to document insertion and query search. `gemini-embedding-001` distinguishes `RETRIEVAL_DOCUMENT` from `RETRIEVAL_QUERY`.
</Warning>

<Note>
  Google requires manual L2 normalization for `gemini-embedding-001` vectors shorter than 3072 dimensions. The example normalizes Agno's 1536-dimensional default. `GeminiEmbedder` returns the provider values unchanged.
</Note>

## Run the Example

<Steps>
  <Snippet file="create-venv-step.mdx" />

  <Step title="Export the API key">
    ```bash theme={null}
    export GOOGLE_API_KEY=your_google_api_key_here
    ```
  </Step>

  <Step title="Install dependencies">
    ```bash theme={null}
    uv pip install -U agno google-genai
    ```
  </Step>

  <Step title="Run the example">
    ```bash theme={null}
    python gemini_embedder.py
    ```
  </Step>
</Steps>

## Developer Resources

* [GeminiEmbedder reference](/reference/knowledge/embedder/gemini)
* [Embedders overview](/knowledge/concepts/embedder/overview)
* [Gemini embeddings guide](https://ai.google.dev/gemini-api/docs/embeddings)