Mats Linder 

The rate limits for the OpenAI API are determined by the type of account you have and the specific endpoint you're using.  They are measured in two ways: requests per minute (RPM) and tokens per minute (TPM).  The rate limit can be hit by either option (RPM or TPM) depending on what occurs first. 

This page from OpenAI might be helpful: https://help.openai.com/en/articles/7416438-rate-limits

 A simple summary of this might be to think about how the plugin works.  It works by sending your text to OpenAI's servers, which then return the translated text based on whatever prompt you created or used.

However, there's a limit to how much text the plugin can send within a certain period of time.  This limit is based not only on the number of requests but also on the total amount of text (or "tokens") being sent.  If the plugin sends too much text too quickly, OpenAI's servers will say, "Hold on, you're asking me to do too much too quickly.  I need a break."  This is when you see the "too many requests" error message.

Now, if you're asking the plugin to do something complex, like creating 30 different translations of a single sentence and making sure they're all no more than 10% longer than the original, this could make the plugin hit the limit faster.  That's because this task requires more processing power and time from OpenAI's servers.

So, if you're seeing the "too many requests" error, it could be because the plugin is trying to translate too much text or perform complex tasks too quickly.  You might need to wait a bit before trying to translate more text or simplify the tasks you're asking the plugin to do.

If waiting or simplifying tasks isn't an option, we could ask OpenAI to increase the limit of how much text the plugin can send within a certain period.  But OpenAI will only agree to this if they believe the plugin really needs it and our users won't abuse it.... and more importantly if doing this fits in with their own work.  Note this statement on the link I provided "Please note that during the limited beta phase of GPT-4 we will be unable to accommodate requests for rate limit increases. In its current state, the model is intended for experimentation and prototyping".

Tagging Patrick Andrew Hartnett for info.

Hi Mats Linder , the model text-davinci-003 is available for free for the first 3 months. You need a paid account to access both gpt-3.5-turbo and gpt-4

You can always check the usage from your account on OpenAI, here: platform.openai.com/.../usage

Thank you Paul and Andrew!

It might be that the problem was that my trial account was no longer valid (the Usage was shown to have expired), although the first API key was created on 16 April. (I cannot find a date for the creation of the trial account.) Anyway, when I signed for a paid account, the app worked. Strangely, though, I always have to Reset Window Layout in order to be able to open it. Not a big issue but mysterious.

So your careful explanations, Paul, were unnecessary in this particular case (sorry to have taken your time once again when there was no need). However, they certainly give more understanding of the mechanisms of this very unusual and interesting feature.

Thanks Mats Linder ... seems I missed the most obvious explanation of all ;-)