Welcome to the AI Models Integration repository! This project provides seamless integration with various state-of-the-art AI models, offering a unified API for efficient interaction.
- Unified API: Interact with multiple AI models through a single, streamlined interface.
- Extensibility: Easily add and integrate new AI models as they become available.
- Scalability: Designed to handle multiple requests efficiently, ensuring optimal performance.
- Rate Limiting: Exponential backoff retries to prevent abuse.
- Caching: Redis is used to cache frequent queries for better performance.
- Validation: All user input is validated with Zod for a secure and reliable API.
- Streaming: Real-time responses for long-running processes.
- Monitoring: Integrated Prometheus metrics to track API usage and errors.
- Anthropic Claude 3.5 Sonnet
- Google Gemini Flash 1.5
- DeepSeek R1
- OpenAI GPT-4o Mini
- Meta LLaMA 3.2 3B Instruct
- Mistral Small
- Node.js: JavaScript runtime environment.
- Express: Fast, unopinionated, minimalist web framework for Node.js.
- OpenAI SDK: Integration with OpenAI's API for AI model interaction.
- Redis: In-memory data structure store used for caching and rate limiting.
- Zod: Input validation for secure and structured API interactions.
- Prometheus: Monitoring of API usage, errors, and performance.
- ioredis: Redis client used for integrating caching and rate limiting features.
git clone https://github.com/JawherKl/ai-models-integration.git
cd ai-models-integrationnpm install- Create a
.envfile in the root directory. - Add your configuration variables (e.g., API keys, model endpoints).
API_KEY=api_key_from_openrouter.ai BASE_URL=https://openrouter.ai/api/v1 REDIS_HOST=127.0.0.1 REDIS_PORT=6379
npm start- The server will run on
http://localhost:3000by default. - Use the provided endpoints to interact with the integrated AI models.
curl -X POST http://localhost:3000/search \
-H "Content-Type: application/json" \
-d '{"model": "change_with_model_example", "query": "Latest advancements in renewable energy 2025"}'curl -X POST http://localhost:3000/analyze \
-H "Content-Type: application/json" \
-d '{"model": "change_with_model_example", "text": "Tesla announced new solar roof technology with 25% improved efficiency in Q4 2023."}'- Rate Limiting: To prevent abuse and ensure fair usage, rate limiting is implemented with exponential backoff. If the user exceeds the allowed number of requests, they will be temporarily blocked from making more requests.
- Caching: Redis is used to cache frequent queries, reducing the load on the AI models and improving the response times.
- Input Validation: We validate all incoming requests using Zod to ensure the data is structured and safe.
- Streaming: For long-running processes, such as AI completions, we stream the responses in real-time.
- Monitoring: The API tracks usage metrics and errors via Prometheus, allowing you to monitor the health and performance of your system in real-time.
Contributions are welcome! If you have suggestions, improvements, or bug fixes, please open an issue or submit a pull request.
This project is licensed under the MIT License. See the LICENSE file for details.
- Special thanks to all contributors and the open-source community.
- Gratitude to the maintainers of the libraries used in this project.
- The Rate Limiting feature uses Redis for storing request counts, ensuring that users cannot flood the system with requests.
- Caching stores frequently requested data in Redis, which minimizes the number of redundant calls to the AI models, improving efficiency and speed.
- Zod ensures that all user inputs are validated before they are processed, making the application more secure and reliable.
- Streaming allows for real-time responses, reducing the wait time for users interacting with models that take longer to process.
- Prometheus provides valuable insights into the health of the API, making it easier to monitor usage, errors, and response times.