I don't know why #Obsidian creators keep recommending Claude as the AI to use to chat with your vault when super cheap but equally good models like Deepseek and Mistral exists.

[liztai@hachyderm.io]

I don't know why #Obsidian creators keep recommending Claude as the AI to use to chat with your vault when super cheap but equally good models like Deepseek and Mistral exists. If you are doing something creative, Claude could be worth it, but for queries and searching knowledge bases, Deepseek and Mistral are already excellent. I was about to get Qwen but it wants my ID! What? Why. So I'll stick to just using their free web version.

Anyway, spent the weekend setting up my TypingMind and getting APIs from Mistral and Deepseek because Openrouter didn't like my credit card. Might as well I guess. I may create a separate Google Drive to copy a vault with files I want TM to query. I would never allow it direct access to my original vault.

I did think of getting something like Claude Code or a CLI front but decided I like a chat interface.

I debated creating a local llm on my private computer but I don't think it's powerful enough for the most basic model.