I have a few services hooked up to using OpenAI and I keep having to update them when you release a new ‘best’ model. What about having a ‘latest’ option that tracks the best model you have at the time (provided the price is not more than the last).
Is this a thing? Would other people find this useful?
Welcome to the Forum!
Personally, it would keep me awake at night to just dynamically switch to whatever is the latest model in a production environment without any further testing 
Thank you!
Yeah, I agree, this would only apply to things like RAG systems etc where you want to keep tracking the best model, but not other systems which need to do things like structured responses etc.
There’s a chatgpt-4o-latest “chatty” model available on the API which will always point to the latest model ChatGPT uses.