task-model-router

What problem does it solve?

It solves the inefficiency and quality loss of using a single LLM for every request by selecting the best-fit model based on the user’s input (chat, code, images, or long documents).

Core Features & Use Cases

• Task classification for model choice: Automatically detects whether the request is vision, reasoning/code, large-context, or general chat and routes accordingly.
• Deterministic routing rules + fallbacks: Uses explicit detection triggers (e.g., image blocks, code signals, token/length thresholds) and falls back to the next best model if the chosen one fails.
• Mixed multi-task handling: Designed for agents that receive mixed content like chat plus code plus images, optimizing both speed and cost.

Quick Start

Ask your Lynn-powered agent to process a mixed request (e.g., “Review this refactor PR and also analyze the screenshot I attached, then summarize the key risks for a long report”), and it will automatically route each part to the most suitable model.