IMatch 2025.4.4 Released - Rate Limits

[Stenis]

As I have written on other things I have observed changed after the last update of iMatch (2025.4.4), when it comes to my personal settings, I now also have noticed that my configuration of Autotaggers "Rate Limits" for my AI models used have changed after installing the new release.

Below you can see the official limitations för these models as displayed on OpenAI "Limits" for the use of their API "Thier One".

Model             Token limits               Request and other limits        Batch queue limits
gpt-4.1              30 000 TPM               500 RPM                                     900 000 TPD

gpt-4.1-mini    200 000 TPM              500 RPM                                   2 000 000 TPD

I had already changed my Autotagger settings to these values in order to get in synch with OpenAI API and get the best possible performance for my API conditions BUT they were reset to the iMatch Defaults for OpenAI API. The impact that had was to severely decrease the processing speed when using Autotagger and it was of that reason I got aware of what had happened.

Now everything is great again and Autotagger is really fast now with Open AI API GPT-4.1-Mini but this was another thing that of some reasons had changed and caused problems.

[Mario]

Rate limits are based on defaults IMatch sets per model. When the user changes rate limits via E > P > AutoTagger, the updated rate limits are stored in the database. Rate limits are maintained per AI / model.

Looking at the code that loads/stores rate limits in the IMatch database, the last change was in January 2025. For the rate limit dialog, the last change was last year.

One of the first things the AI engine in IMatch does is loading the stored settings from the database, including the rate limits.
Looking at my test databases, they all have the correct (Tier-1) rate limits I've dialed in, and these databases are older than a year and have been used during AutoTagger development.

I don't really see why only the rate limits on your PC can have changed after the update. As I said, the code is old and untouched for many versions. No changes to the setting object were done for months.

@Other Users who use OpenA: 

Can you please check the Rate Limits and see if there are as you have set them.
If OpenAI rare limits are default (and thus severely liming) the AutoTagger configuration dialog shows a corresponding message, asking you to set the rate limits to match your OpenAi usage tier.

[Stenis]

Strange