[AIT] is required for any discussion with an AI topic, such as this one. Please update your title.

I've used Fabric to summarize the transcriptions. I believe it works with Ollama and local models.

Just forget about the videos you want to summarize and (optionally) eat a cheeseburger instead. I came up with this solution myself. It works instantly, with 0 dependencies, and has a 0% fail rate. It is my intellectual property though, so please pay me $19.99 per day until you get tired of not summarizing youtube videos. Cheeseburgers are not included, btw.

AI summary of AI slop for a delicious concentrated slop juice

https://github.com/nodiscc/toolbox/blob/master/SCRIPTS/llamacpp-summarize

Kagi Summarizer (not FOSS, though).