I have built an app that does summarization outputs of an ongoing flow of source documents using 4o. I have been testing 4.1 and the content of the resulting summaries is clearly better but the output is excessively long. Like 3x or 4x longer than what 4o produces. This is too long to be useful for our application.
I don’t want to just cut off the output with post-processing. Has anyone else experienced this? Does anyone have any techniques or ideas for encouraging shorter outputs from 4.1?