OpenAI made some changes to ChatGPT that make Custom Instructions infinitely more powerful. I spent six hours trying to get a Chat GPT to understand one of my instructions. This will now fix that.
You can choose which model your Custom GPT’s run under. Instead of defaulting to GPT4o, you can now choose a reasoning model. These models pay better attention to your instructions. They can even use tooling when necessary.
When to use this
Choose this when you create a new Custom GPT, or modify it from the GPT menu. Users can also override this setting when they use Custom GPTs.
Huge improvement! Nice job ChatGPT, they’re now releasing useful things like this every few days.
