Assistant settings in dashboard vs API

When I configure an assistant’s settings (like temperature, top_p, and model) on the dashboard at https://platform.openai.com/assistants/ … do I need to redefine those same parameters when I want to use the assistant via API calls?

Or is it set-and-forget. So for example, I set temperature via dashboard = 0.8. Then I use the same assistant ID in a thread via API call in some coded application. Will that thread assistant inherently have a temperature of 0.8?

1 Like

Good question. Might be a good idea to implicitly set it, but you could test.

If you do, let us know! :slight_smile:

It’s set but you CAN provide both explicitly in a new run which will override the default.

@user978 - Yes, when you create an assistant, it is bundled together with all the configurations. You can retrieve the assistant ID to access the assistant, which includes instructions, tools, temperature settings, and more. However, you can explicitly modify these according to fit your specific use case.