the list is pretty long

I use: ChatGPT/Claude/Bard for brainstorming/random questions or explorations. There are several reasons I use multiple llms at once (Bard by far the least). One is that I've had mixed experience with their creativity in problem solving (for example i used the same question which was how to solve a decentralized reputation system that is entirely public key cryptography based on most LLMs out there and the results are wildly different, from not even understanding the question properly (Bard) to coming up with decent strategies (Claude). The other reason is multitasking (as i'm too lazy and cheap to run multiple api based chatgpts locally). I also used pdf reading frequently to summarize things and evaluate if i should actually read the documents.

GH copilot, Cursor and chatgpt for coding - mostly i treat chatgpt as a junior dev that works for me, specially where i'm rusty with the syntax but know exactly what i want to achieve (which describes majority of my coding)

I have several things running "locally", whisper for my transciber bot, ollama for testing various llms locally, SD for photo to text experimentations

Dalle, Midjourney, and occasionaly SD for brainstorming, memes etc