I’m experiencing issues with the consistency of outputs from my OpenAI assistant. I’ve created an assistant using the 4o-mini model and set up a system prompt to analyze a PDF and summarize it. When I test this in the Playground, the output is as expected and works perfectly.
However, when I run it through Make. com using the OpenAI Message Assistant module, the responses become inconsistent. Sometimes it can’t read the PDF or extract data, while other times it works perfectly. Additionally, there are instances where the output is missing some data or doesn’t include all the expected information.
Has anyone encountered similar issues or have suggestions on how to improve the consistency of outputs? Any help would be greatly appreciated!