MultiTurnRoute
An LLM routing platform for multi-turn dialogue scenarios that provides a shared interface, the Chain-of-Route routing algorithm and an Arena evaluation mechanism, aiming to optimize model selection and cost efficiency in multi-turn conversations.
Timeline (last 90 days)
MultiTurnRoute is a multi-turn dialogue LLM routing platform that provides a shared interface, the Chain-of-Route algorithm, and an Arena evaluation mechanism.
MultiTurnRoute's shared interface gives built-in routers and user-defined routers three types of information: the current query, the conversation history, and an estimate of each model's reusable prefix length.
On this benchmark, when reaching the quality level of the strongest single model, CoR reduces normalized cost by up to 20.3% relative to the strongest routing baseline, while peak reply quality is on par or improved.
The platform is currently based on annotated data from five LLMs, and the router's generalization ability on larger model pools or conversations in different domains remains to be verified.
All Facts (4)
MultiTurnRoute is a multi-turn dialogue LLM routing platform that provides a shared interface, the Chain-of-Route algorithm, and an Arena evaluation mechanism.
50%UnverifiedMultiTurnRoute's shared interface gives built-in routers and user-defined routers three types of information: the current query, the conversation history, and an estimate of each model's reusable prefix length.
50%UnverifiedOn this benchmark, when reaching the quality level of the strongest single model, CoR reduces normalized cost by up to 20.3% relative to the strongest routing baseline, while peak reply quality is on par or improved.
50%UnverifiedThe platform is currently based on annotated data from five LLMs, and the router's generalization ability on larger model pools or conversations in different domains remains to be verified.
50%