Add provider-specific model lists
Resolved 💬 5 comments Opened Oct 31, 2025 by nihaals Closed Dec 31, 2025
💡 Likely answer: A maintainer (etraut-openai, contributor)
responded on this thread — see the highlighted reply below.
What feature would you like to see?
The model selector always shows gpt-5 and gpt-5-codex. It would be helpful if you could define your own list and scope it to a specific provider to support both wanting to use an OpenAI model other than GPT-5 or using a provider with models/model names different to OpenAI's.
Additional information
See #4283.
5 Comments
This feature request hasn't received any upvotes in over two months, so I'm going to close it.
It is sad that this feature request has been closed as not planned. Everybody who serves gpt-oss-120b (-20b) locally as a not so strong but alternative option for data privacy reasons under certain circumstances runs into trouble.
I do so with the llama.cpp llama-server. So still all within the OpenAI world. And disappointing that we get into trouble using the /model command, which offers to switch to models which are not available from that local provider. Using Codex to implement such a feature would not be too much effort, I would guess. If I were able from my js coding skills (I do not have any) to do so, I would create a PR. But I would have to rely 100% on Codex. I doubt such an amateur - PR would be welcome.
P.S.: You drive your own ChatGPT into difficulties. I asked for a solution for my trouble and it answered only with this closed feature request or semi useful workarounds.
@gsgxnet, as I noted, this feature request didn't resonate with other Codex users as evidenced by lack of upvotes. It may have been the way it was described or framed. If you'd like to open a new feature request with a different description or framing, you're welcome to do so.
@etraut-openai - There are already similar requests: https://github.com/openai/codex/issues/8319
Looking thru the requests-issues here it seems that instead of upvoting it happens a few times that the same features are requested again.
That 8319 request is not mentioning llama-server. I will add
llama-serverthere.We try to detect duplicate or overlapping feature request and consolidate them, but sometimes we (Codex and I) don't catch duplicates.