I understand your question. Unfortunately, using your own API key won’t significantly address this issue. It doesn’t stop you from using it, but it won’t solve this particular problem.

The only thing that can speed up response generation is reducing the question’s size. In the widget settings, under the GPT tab, there’s an option called “Document,” which determines how many parts of your resource are included in the context for generating the answer.

Reducing this value will affect generation speed. Also, turning off additional Prompt settings will make response generation a bit faster, but not significantly.

Soon, in OmniMind, the GPT-4-Turbo model will become available, and with it, your Omni-bot’s work will be several times faster. That might be the best solution to this problem.