RouteLLM has received an upgrade, enhancing its ability to select the most suitable language model (LLM) based on user prompts. The updated system can now effectively route complex queries to the o1 model, quick answers to gpt4o, coding tasks to sonnet, simple code requests to deepseek, and long context inquiries to gemini. Additionally, a new Router Agent Workflow has been introduced, which automatically matches tasks to specialized models and provides clear justifications for its decisions. This workflow is powered by TogetherCompute and includes a simple API for executing the chosen model. The smart routing system employs a weak supervision approach to estimate model quality through output comparisons, allowing it to select the best LLM without needing labeled data.
RouteLLM is insanely good at picking the right model based on the context of the query. Easily one of my favorite features of ChatLLM, it understands the query and routes perfectly: - o1: Complex queries - gpt4o: Quick answers - sonnet: Coding - deepseek: Simple code - gemini:… https://t.co/2bUKsZxleb
Smart routing system that knows which LLM to use just by looking at their outputs A method to select the best LLM for different inputs without needing labeled data, using a weak supervision approach that estimates model quality through output comparisons. ----- 🤔 Original… https://t.co/HTuE01xt8B
New Cookbook: Router Agent Workflow! Learn how to use an LLM as a decision-making router! We implement: 1. LLM Router: A language model that selects and justifies the best model for a given task 2. Simple API that executes the chosen model and solves the task https://t.co/jtNbFTJ27z