Cursor Unveils 'Cursor Router' to Cut AI Costs by 60% with Automatic Model Routing

Cursor, an AI coding platform, has officially launched its model routing system 'Cursor Router' to significantly reduce enterprise AI operational costs. The AI classifier, trained on over 600,000 real user requests, analyzes task difficulty and context in real time to automatically select the most suitable AI model. By prioritizing cost efficiency, Cursor Router addresses the issue of developers using a single high-performance model for all tasks, which leads to excessive AI cost growth. The system optimizes performance through A/B testing with millions of service requests and considers factors like task complexity and domain characteristics to assign models. It also accounts for 'prompt cache miss' costs, ensuring accurate cost savings projections. The router dynamically updates routing policies to adapt to evolving AI ecosystems.