If you’ve been experimenting with custom GPTs inside ChatGPT, here’s a welcome upgrade: OpenAI now lets users pick from all available models when building or running a custom GPT, not just GPT-4o. Here’s how it works (and what’s missing).
What are custom GPTs again?
If you have no idea what this is, you’re not alone. OpenAI launched custom GPTs in late 2023, but the feature hasn’t exactly taken off the way the company probably hoped. Here’s how OpenAI describes them:
“…custom versions of ChatGPT that combine instructions, extra knowledge, and any combination of skills.”
In a nutshell, suppose you always use ChatGPT to help with social media SEO. Instead of retyping your instructions every time, you can create a custom GPT with saved prompts, settings, behaviors, and even files, then just start from there whenever you need it.
I’ve made custom GPTs to help my girlfriend study for tests, for instance. At Gigahertz, me and Gui Rambo even have one for quickly pulling info from our episode transcripts. It’s surprisingly handy. There’s even a sort of app store filled with user-created GPTs.
So what’s new?
Up until now, users could only build custom GPTs with GPT-4o. But now, you can choose between GPT-4o, o3, o4-mini, o4-mini-high, GPT-4.5, GPT-4.1, and GPT-4.1-mini.
Yes, OpenAI’s model naming situation has gotten messy at best. Which is why they also added a “Recommended Model” setting, giving GPT creators a default starting point, while still letting users switch models on the fly. As OpenAI explains it:
“When the user loads a new conversation with the GPT the recommended model will be used. If the model is not available to the user, a similar model may be automatically selected.”
... continue reading