Hello,
I am building a workflow and I’ve noticed that my ‘Generate Text’ blocks are sending the entire chat history in every request to the AI model.
In my recent debuggers, the prompt size reached over 37,000 tokens, even though my System Prompt is much smaller. This is significantly increasing my token usage and costs.
Could you please guide me on how to:
-
Disable the automatic inclusion of chatHistory in a specific ‘Generate Text’ block?
-
Configure the block to only send the current System Prompt and specific variables without the previous messages from the session?
I want to ensure each run is independent to optimize performance and reduce costs.

