I'm curious about the specific models that power GPTs. A lot of sources online claim they run on model 4o, but I haven't seen any official confirmation from OpenAI. I've been experimenting with my own GPT and noticed that in the iPhone app, if I have a message from the GPT, I can tap its name at the top, go to "see details," and check the "Model info." So far, I've come across models o3, 4o, and even 4.5. Each time, it says underneath, "Newest and most advanced model." I'm starting to think that the model might change based on the context—like switching to 4o for multimodal responses and o3 for reasoning tasks. Has anyone else noticed this? Any documentation I might have missed?
2 Answers
You're right about the naming confusion. When they say 'GPT,' they could mean a whole range of things like a standard ChatGPT, or the specific tool with a custom prompt. The app allows you to choose various tools like 'Data Analyst' or even create new ones to share, which adds to the mix. I think terminology from OpenAI could be clearer—it’s a bit messy, honestly.
From what I've seen recently, it seems like the GPTs are mostly running on chatgpt-4o-latest now. There’s a lot of speculation since not much official info is shared. But I've noticed in the app, it often shows these tags under each interaction. Curious to hear where you got your info from, though. Seems like there might be bugs or inconsistencies in how the model info is displayed!
I grabbed my info straight from the app itself! Here’s a screenshot showing my exchange with the GPT. I think there could be issues with how the app reports the models—just seems a bit off! Despite that, I'm with you on this; it feels like a mix-up or maybe even an A/B test!
Absolutely! It's confusing when people use 'GPT' to mean different things. I keep seeing folks mix up the models with the custom tools built off them. It definitely feels like OpenAI needs to clarify their naming conventions!