https://github.com/erickrus/google_genai_to_ollama_service
converting Google’s GenAI API (e.g., Gemini models) into an Ollama-compatible API interface
https://github.com/erickrus/google_genai_to_ollama_service
genai ollama proxy
Last synced: 6 months ago
JSON representation
converting Google’s GenAI API (e.g., Gemini models) into an Ollama-compatible API interface
- Host: GitHub
- URL: https://github.com/erickrus/google_genai_to_ollama_service
- Owner: Erickrus
- Created: 2025-04-01T03:34:19.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2025-04-01T04:01:16.000Z (6 months ago)
- Last Synced: 2025-04-01T04:28:01.997Z (6 months ago)
- Topics: genai, ollama, proxy
- Language: Python
- Homepage:
- Size: 0 Bytes
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# google_genai_to_ollama_service
## 1. What is the Service?
This service acts as a proxy, converting Google’s GenAI API (e.g., Gemini models) into an Ollama-compatible API interface. It bridges Google’s generative AI capabilities with Ollama’s local LLM ecosystem, supporting both streaming (real-time output) and non-streaming (full response) modes. This allows tools or clients expecting an Ollama API to leverage Google’s cloud-based models seamlessly. Following APIs are implemented.
- POST /api/generate
- POST /api/chat
- POST /api/embeddings
- GET /api/tags
- GET /api/version
## 2. How to Run
### Requirements
- Python 3.12+
- Google API key (from Google AI Studio or Vertex AI)
- Install dependencies:
```shell
pip install langchain langchain-google-genai flask python-dotenv
### Setup
Set API Key: Create a .env file:
```
GOOGLE_API_KEY=your-google-api-key
```
### Run the Service:
```shell
python3 google_genai_to_ollama_service.py
```
Expose (Optional): Use a tunnel service like ngrok to make it accessible widely:
```shell
ngrok http 11434
```
Copy the ngrok URL (e.g., `https://something.ngrok-free.app`) for external access.
### Notes
- **Port**: Uses 11434 to match Ollama’s default port.
- **API Compatibility**: The script simplifies the Ollama API format; adjust endpoints or response structure if your Ollama client expects more fields (e.g., `model`, `created_at`).
- **Model**: Uses `gemini-2.0-flash`; update to your desired Gemini model.