### Answer Based on the available sources, I don't have specific information about the LLM Playground version-switching behavior you're experiencing when clicking "new" after publishing updates. The LLM Playground allows you to test agents in chat for rapid prototyping and debugging [(1) ]() . When you access it from your agent's detail page under the "Test LLM" tab and choose "Manual Chat," you can test conversations, observe agent responses, and iterate on your configuration . To reset conversations in the playground, you can click the "delete" button . You can also save test cases by clicking the "Save" button and adding a descriptive name, which can be accessed later from the agent detail page . However, the specific issue of version numbers reverting when starting a new test isn't directly addressed in the documentation. This could be a UI behavior or caching issue. For testing with specific versions, the API supports a

version

parameter that accepts numeric versions, "latest", "latest_published", or tag/environment names [(2) ]() . I'd recommend reaching out to Retell support about this specific version-switching behavior, as it may be a bug or require clarification on the intended workflow. Visit [Documentation]() or [Hire certified Retell partner]() for additional assistance.

​ To learn more, see the sources I considered: [[1] Manually test your agent](https://docs.retellai.com/test/llm-playground) [[2] Agent Playground Completion](https://docs.retellai.com/api-references/agent-playground-completion) ​

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