Dear OpenAI,
While on the whole I am very happy with your service level and tools. I do have a couple of bug fix requests.
It’s the little things!
Now we have put some pretty serious thought into this and it appears to be impossible with Chrome to use ChatGPT without a 
. It would be wonderful if we could find some solutions to these issues. Edge does not have this issue and Chrome never used to. Neither browsers now Auto Focus the User chat box on load. We are also being forced to ask overly complex questions.
I can’t find fixes to these issues at least not within the OpenAI Ecosystem.
Issue 1: Smart/Default Model Selection Setting
Settings on ChatGPT should include the ability to set the default model on site load.
Defaulting to last used model is not an ‘intelligent’ setup.
It is incredibly frustrating to loose 10% of my 50 weekly o1 searches to quick queries that would be just as easily serviced with 4o. It’s not about the money, it’s about the needless waste and resulting lack of intelligence.
While I, as a programmer, can use the API to ‘top-up’ I understand that the majority of users don’t have that option.
Unless specifically set it not only inconveniences smaller users with low limits, it is impacting everyone with rate limit reductions/speed.
For anyone else grappling with this bug there seems to be no way around this… setting a bookmark with model specified also doesn’t work if not using the dropdown model selection on chatgpt.com.
A ‘smart’ option might be an AUTO-SELECTION of model default based on the query if not specified by the user… If the query was for a simple translation and the model wasn’t specifically set then the ChatGPT request model would be down-graded automatically from o1 to 4o or 4o-mini.
Issue 2: Input Selection Focus
Further the chat window (at least on Chrome and it appears Edge too) no longer auto-focuses the text input box on load like other search engines/chat applications, you have to use the .
This is excessively complicated in this day and age, it feels like I’m back on Windows 3.1.
Summary
Issue 1: Smart/Default Model Selection Setting - Seems like a ‘no-brainer’.
Issue 2: Input Selection Focus - The issues seemed to start around the time o1/Projects was introduced. I think maybe Projects loads after chat box so you need to refocus the chatbox after.
If these issues are intentional design choices aimed at ‘Throttling’ users then this raises concerns about fairness, adding a 1 second pause after clicking search but before every search processed would be far less stressful for users than reaching for the mouse and clicking the search box (especially ‘Power Users’)… (I have put this through ChatGPT for training purposes)
Sincerely,