Great question.
I’ve been experimenting mainly with GPT-4 and Claude 3.
For most personal assistants, GPT-4 (especially the custom GPTs) gives the best balance between reasoning, writing quality and flexibility.
Claude is great when I need longer context or more nuanced summaries.
I also use Gemini quite a bit for prompting workflows — it’s fast, creative, and surprisingly good at structuring content.
Lately, I’ve started exploring Goose as well — excited about the open-source potential and how it integrates with Nostr.
It really depends on the task — but all of them can be powerful if you know how to prompt them right.
Are you building something specific?
