Evals is a framework for evaluating OpenAI models and an open-source registry of benchmarks.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).