the thing chews through claude usage like a rabid dog. i've not figured out what model to run it with to keep it cheap but still useful

Yeah it's a little weird how the token usage is so high after just a few prompts.

I'm also using it with the z.ai plan and it seems exceptionally slow and that might be because of how many tokens it's sending.

That being said, one I got speech to text setup that did work decently.

Also, scheduling a reminder with it wasn't successful. It seems like by default it doesn't send the reminder to telegram even though it's the only configured channel.