Our platform not only evaluates models but also helps partners select the best-performing models based on user feedback.

We have developed an internal evaluation system to measure our models, as relying on external benchmarks often leads to data contamination and unreliable results.

As we gather more data and build a larger platform, we can create more granular and personalized evaluations of AI models.

Cursor prioritizes using the best pre-trained open-source models and collaborates with model providers to enhance their capabilities without reinventing the wheel.

We aim to provide a platform that allows users to build on our models, offering tools for customization and control to create effective products.

We work with both large and small model providers, aiming to assist anyone who wants to collaborate with us on AI evaluation.

We work with both large and small model providers, aiming to assist anyone who wants to collaborate with us on AI evaluation.

The 'prompt to leaderboard' approach teaches us that we can evaluate models by learning how they perform in various scenarios, rather than relying solely on static benchmarks.

As an application layer company, we must adapt to the rapidly changing landscape of AI model providers and ensure we are testing and integrating the best models available.