Silene Bench

SileneBench offers a unique experimental platform to explore long-term primary care management. Key features include: * Recurring patient interactions * Complex disease progression over time * Limited resource allocation challenges * Delayed consequence modeling * Capacity planning and operational decisions This playable benchmark simulates a primary care clinic environment, challenging users to manage patients who return over an extended period. It incorporates the complexities of diagnosing rare diseases amidst common ailments and requires careful consideration of resource limitations. Users can actively run the clinic, making both clinical and operational choices, or observe advanced models like GPT-6 Astra navigate these scenarios. The platform is designed as a foundational step towards a year-long simulation, providing a robust environment for evaluating decision-making processes in a dynamic healthcare setting. Future updates will allow developers to connect their own agent systems for comparative analysis, fostering advancements in systems designed for complex, long-horizon operational management.