Mistral Embedder

Generate 1024-dimensional embeddings with MistralEmbedder and mistral-embed.

MistralEmbedder uses mistral-embed by default. The model returns 1024-dimensional vectors.

from agno.knowledge.embedder.mistral import MistralEmbedder

embedder = MistralEmbedder()
embedding = embedder.get_embedding("The quick brown fox jumps over the lazy dog.")

print(embedding[:5])
print(len(embedding))

In the current adapter, endpoint and max_retries use parameter names rejected by the current mistralai client. Pass a preconfigured client through mistral_client when you need either setting.

Run the Example

Set up your virtual environment

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

Export the API key

export MISTRAL_API_KEY=your_mistral_api_key_here

Install dependencies

uv pip install -U agno mistralai

Run the example

python mistral_embedder.py

Developer Resources