Issue: As of the last week or so, I am no longer able to select and use the o1 model after running a web search at any point earlier in the same chat thread.
For context:
- I am not trying to actually run a web search with o1, as o1 doesn’t support web search functionality.
- Instead, I am trying to continue a conversation with o1 after retrieving some recent information from the web (using either 4o or o3-mini).
- The issue persists even after clearing all chatgpt cookies and stored page information, signing out then signing back into my account, and doing a “hard page refresh” which clears cache and re-retrieves all of the webpage’s assets and files.
This was possible up until recently. For example, I occasionally use this workflow to do an initial dive into a new API or software development tool I’m unfamiliar with. Being able to use web search to retrieve up-to-date docs then have o1 reason through some of the finer details is very powerful.
NOTE that I am still able to successfully select and use o1 after a web search if I manually change the model
query parameter in the page URL to model=o1
. (See screenshot in the reply below).
This issue is also present in the IOS app version 1.2025.021 (12961324711)
:
It’s possible that this is intended behavior since both the web and mobile app frontends enforce the same restriction, but I don’t think that’s the case. To me, blocking such a useful feature (which still works, in the web version, with the workaround) is definitely a bug.
- If anyone has a better workaround than manually changing the
model
query param, please let me know. - Otherwise, a fix would be much appreciated!