Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer When using the Retell SDK, the version number defaults to the latest version of the agent if not specified. [(1) ]() For example:

python
get_agent(version=2)  # Gets the version with version number 2.
get_agent()           # Gets the latest version of the agent.

So if you run tests without specifying a version, it will use the latest version. After publishing a new version, subsequent test runs should automatically use that newly published version—no need to manually specify it unless you want to test against a specific older version. If you want to lock tests to a particular version, you can explicitly pass the version parameter.

​ To learn more, see the sources I considered: [[1] Setup versioning for agents](https://docs.retellai.com/agent/version) ​

### Answer Based on the available sources, I can share what's documented about evaluations (evals) in the Retell AI UI: In the AI QA section of the dashboard, the Detailed Calls tab shows an Eval column for each analyzed call. [(1) ]() When you click on a call row, you can view the Call QA Sheet which displays: - **Overall Score**: Pass/Fail status with numerical score - **Passed Metrics**: Metrics that met defined thresholds (shown with green checkmarks) - **Failed Metrics**: Metrics that didn't meet thresholds (shown with orange warning triangles) You can also calibrate individual call QA results by manually adjusting metric evaluations—marking passed metrics as failed or vice versa—and add custom notes to any call QA. However, the sources I have don't specifically address whether running tests in the UI automatically picks up newly published agent versions versus requiring manual specification. The documentation I have access to covers the QA results viewing and calibration features, but not the version behavior for simulation testing in the UI. Would you like me to try to answer generally based on my knowledge, or would you prefer to check the official docs or reach out to support for clarification on version handling in UI-based evals?

​ To learn more, see the sources I considered: [[1] View QA Results](https://docs.retellai.com/ai-qa/view-qa-results) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.