One of the worst experineces I have had with AI to date.

One of the worst experineces I have had with AI to date. Your token usage in the past 3 days has been insane compared to Codex. I spent over half my month's alotment inthe first two datys just trying to get things setup to work against my Obsidian vault and LLM Wiki. Once I got throug the initial setup, I had used half my monthly alotment of tokens. I tried doing some work on my writing projects but at your usage rates, I would at least ahve to upgrade to your $40 model to even copme close to matching what I get for $20 at OpenAI or Anthropic. THere is no value here. I'm going to go try OpenRouter next with OpeWebUI and see if I can make that more cost effective. After this experience, I think I'll stop using GitHub AI crap at work as well and focus on the other tools. I might abandon GitHub for code repository as well. ALl of this just feels very Microsofty and I don't like it. I don't want to play those games any more.

No source file, test, or specific technical change is identified. Start by reproducing the setup and chat workflow at /chat on app version 1.1.20 under Windows, then measure token usage against the reported sequence. Done would require a confirmed usage problem and a clearly agreed scope for addressing it.