Evaluations are a crucial aspect of building reliable and high-quality Large Language Model (LLM) applications. The new Pytest and Vitest/Jest integrations with LangSmith provide a flexible, familiar interface for running evaluations, allowing developers to assess performance and ensure quality consistency as they make updates. These integrations offer benefits such as debug capabilities, logging metrics beyond pass/fail results, sharing results with teams, built-in evaluation functions, and real-time local feedback. They also integrate well with existing testing frameworks, providing a seamless experience for developers. With these new integrations, developers can easily track test cases, log inputs and outputs, and receive rapid feedback on their tests, making it easier to spot and fix issues as they go. The approach is more flexible and intuitive than traditional evaluate() methods, allowing developers to define specific evaluation logic for each test case and providing real-time local feedback during testing.