Any error messages in the logs?

When you say it stopped working, that gives me absolutely zero to work with. What exactly isn't working? Are you getting any error messages? What is the AI returning, if anything?

Alex asked about the logs too. Help us help you - give us the specific steps you're taking and the exact result you're getting.

Richard   People are used to saying this in real life.  In my old job, where I was the IT guy, my co-workers said this all this time.  But I could read their minds somewhat because I was familiar with their routines, knowledge level, working environment, and what was in their PCs (which I built for them).  Hence, hearing "doesn't work" was often all I needed to go on.  But of course, that wouldn't work with total strangers on the Internet.

Yep, Kevin, that's exactly it. When you know someone's setup and routine, "it doesn't work" can sometimes be enough to get you started. With a stranger on the Internet, though, there are about 5 million possible causes.

Rusty, when you click Send to OpenAI, what exactly happens? Does it return anything at all? Do you get an error message? If so, put the exact message in a MessageBox or copy it here so I can see it.

I need to know what it did before, what it isn't doing now, and what changed - software, hardware, API key, anything. It's like going to the doctor and saying, "My arm hurts when I lift it this way." Without some more information - and maybe an X-ray - there's no way to tell what's wrong.

"Don't lift it that way." :)

In a nutshell, I'm not aware of any changes OpenAI has made that would invalidate the API code I've been using since I started these Access/OpenAI videos. Other than changing the model from GPT-4 to GPT-5 and now GPT-6, I haven't had to change anything, and the code still works.

They've added a few optional settings over time, like temperature and other controls for response randomness, but the core API interaction has been pretty much the same from the beginning.

So the next step is still to find out exactly what response or error you're getting. That will tell us whether this is an API issue, an account/billing issue, a model issue, or something local in the database.

Richard  Maybe AI can help you with this.  Maybe it can detect lack of detail in a message and send an auto-reply asking for more info, saving you the time for manual replies.  This kind of messages will not stop, Richard, because it's a part of human nature.

OK, I finally figured it out. I was still using the older model. Changed it to: Const MODEL = "gpt-4.1" and everything works as it did before. Thanks very much. I have one client that English is a second language and she relies on this program to reformat her statements. One of the best routines I have ever gotten from you. Thanks very much.

Yeah, the one good thing I've learned about AI is that AI is very good about helping you with AI. If you have problems with your OpenAI code, just ask ChatGPT in the browser. It knows its own software really well.