To each their own, how are you using these extra features? I personally am not looking to spend a bunch on API credits and don't have the hardware to run models larger than 7-8b parameters. I use local llms almost exclusively for formatting notes and as a reading assistant/summarizer and therefor don't need these features.