Early on in my AI adoption journey, the first move I made was to try the free tier of ChatGPT. I was quickly convinced by a friend to upgrade to the Pro subscription for $20/mo so I could access the newer models being introduced at the time.

I kept that ChatGPT Pro subscription for over two years. It’s been good to me. We were good friends.

However, during one of my Personal Retrospective sessions, I realized a few things about my usage:

  • I only used ChatGPT for personal projects, meaning my usage was limited to nights and weekends. At work we have our own enterprise subscriptions
  • My patterns were highly sporadic. I’d have deep, back-and-forth conversations one day, and then go full weeks where I’d only ask a couple of relatively simple questions.
  • I had never tried any non-OpenAI models for tasks outside of programming.

This led me to do some research, and I discovered TypingMind, a frontend web interface (and mobile app) that lets you bring your own API keys for any model. It also helped with timing that I discovered this around Thanksgiving when they were running a 60% off sale on their premium lifetime license.

Now that I use TypingMind and pay strictly for API token usage, my average bill for random ad hoc chats has dropped from a flat 2 a month.