OpenAI's usage thresholds for switching between ChatGPT models or initiating cooldown periods

[hamza.zerarka]

I’ve been exploring the capabilities of the ChatGPT models and am intrigued by the operational intricacies of these powerful tools. I understand that to manage resource allocation and ensure quality service for all users, certain mechanisms like model switching (e.g., from ChatGPT-4 to ChatGPT-3.5) and cooldown periods might be in place, especially under heavy load conditions or after extensive use.

Could anyone provide insights or direct me to documentation regarding the specific usage thresholds that trigger these mechanisms? Thank you in advance for your guidance and support!

[Diet]

Welcome to the community!

@hamza.zerarka

Could anyone provide insights or direct me to documentation regarding the specific usage thresholds that trigger these mechanisms?

If you go on chat.openai.com and click on the model selector, you’ll get this:

Unfortunately, some users have sometimes reported lower thresholds (as low as 20) - so this is not super reliable information and generally not very well documented

[hamza.zerarka]

Thank you, unfortunately it’s not documented