Hybrid AI Router

Hybrid AI Router is a lightweight AI routing system built to improve the efficiency of language model inference. Instead of sending every request to a cloud model, it first analyzes the prompt by identifying its type and estimating its complexity. Based on this analysis, the router decides whether the task can be handled by a local language model or should be sent to a Fireworks AI hosted model. Simple tasks such as factual questions, summarization, sentiment analysis, and named entity recognition are processed locally whenever possible, helping reduce cloud token usage and response costs. More complex tasks like code generation, debugging, logical reasoning, and mathematical reasoning are routed to a stronger cloud model for better accuracy. The project also tracks token usage and latency to measure performance. It is fully containerized using Docker, making it easy to deploy and reproduce in different environments. The goal of the project is to provide a practical and cost-efficient way to combine local and cloud AI models without sacrificing answer quality.