Experimenting with vision capabilities in OpenAI's gpt-4o

I just gave the new gpt-4o AI model a try in SketchUp. One interesting aspect is that you can use it to analyze an image, e.g. a screenshot. Feel free to give it a try using the snippet on my following post. I am curious what people would use something like this for.


Also, gpt-4o works with my OpenAI Explorer (Experimental) extension, albeit without vision capability (so far).

Edit: I should have included an image example. Here is one:


A quick follow-up: I just implemented this in my OpenAI Explorer extension…

I gave this a quick try when you posted it earlier… silly question: do we need to have tokens available? It seems when I tried your Explorer previously I had tokens or something has changed with project api_keys (or both). I think this may be why it didn’t work for me. Otherwise, I’ll post the errors I received to track down the issue.

Any old OpenAI API key should still work (although you can always generate a new one easily). But for tokens, OpenAI gives you a free contingency whereafter you have to pay.

What I have done is paid a $5 credit to OpenAI, after which they just take off any costs from that amount. Because the requests are so cheap, I have still to use that amount up.

Thank you. I did generate a new API key as I was trying to get up and running. It seems the included tokens expire… yours was the only app I spent them on and it was only run a few times. Missed opportunity on my part!

I’m going to follow your suggestion and purchase more so I can try the app.