Python Ollama embeddings with Poetry

Embeddings example

This example shows the smallest OpenAI Python SDK setup for calling POST /v1/embeddings through LunarGate into a local Ollama model.

Source code: github.com/lunargate-ai/gateway-examples/tree/main/python-ollama-embeddings-poetry

Use this example when:

• you want the smallest possible embeddings example
• you want to test local Ollama with LunarGate before building RAG
• you want to confirm that your gateway exposes /v1/embeddings

Need chat plus retrieval?
Move to the Streamlit RAG example when you want uploads, vector search, and answers grounded in local files.

Need the smallest chat example first?
Use the basic Poetry example if you want to verify the standard chat path before embeddings.

What it demonstrates

• OpenAI(base_url=...) pointed at LunarGate
• one client.embeddings.create(...) request
• local Ollama as the upstream embedding provider
• a minimal route matching /v1/embeddings

Run it

cp .env.example .env
poetry install

Make sure Ollama has the embedding model:

ollama pull nomic-embed-text-v2-moe

Start the gateway in a separate terminal:

cp config-simple.yaml.example config.yaml
lunargate --config ./config.yaml

Then run the client:

poetry run python main.py

What to inspect

• main.py for the OpenAI SDK embeddings call
• config-simple.yaml.example for the /v1/embeddings route
• README.md for local Ollama requirements

Why this example matters

This is the shortest path from:

• local embeddings in Ollama

To:

• OpenAI-compatible embeddings through LunarGate

That makes it the right smoke test before you build semantic search, retrieval, or RAG on top.

Good companion pages

• Read HTTP API for the public embeddings endpoint.
• Read routing if you want separate chat and embeddings routes.