According to CoinMeta, OpenRouter has launched Ori Eval to help developers determine which model to use for their AI applications. This tool scans the codebase to find where models are called and tests multiple candidate models with actual tasks within the project. Developers can specify whether they prioritize performance, speed, or cost. During the evaluation period, Ori Eval locks in the testing framework, model configuration, and inference intensity, and finally provides a ranking to avoid comparisons between different models due to varying test conditions. In addition, it also checks whether agent is using the correct tools to prevent unnecessary operations. Developers describe a bug in natural language, and Ori Eval can generate tests that reproduce the issue. After repairs, these tests are integrated into GitHub Actions, and subsequent changes to models, prompt, or code can automatically check if the problem reoccurs. Public rankings can only compare the general capabilities of models, while Ori Eval transforms the manual process of testing models one by one into an automated selection process based on real business needs.