OpenAI
OpenAI is the standard hosted provider flow for many RealTimeX installations.
Open it from Settings > AI Providers > LLM, then choose OpenAI.
What you need
You need:
- a valid OpenAI API key
- a working OpenAI account with billing or quota that allows model access
The current RealTimeX page loads the available model list from your API key. If the key is invalid or the account cannot access models, the model selector stays empty or fails to load.
Current setup flow
- Open
Settings > AI Providers > LLM. - Choose
OpenAI. - Paste the API key.
- Wait for RealTimeX to load the available chat models.
- Choose the model you want as the system default.
- Save the settings.
- Test a normal chat.
How model selection works
The current OpenAI setup is dynamic.
RealTimeX calls the OpenAI model list for the supplied key and builds the dropdown from the result.
Important behavior:
- model options can change with the account behind the key
- the list loads only after RealTimeX has a usable key
- models are grouped by organization in the selector when OpenAI returns organization metadata
This means the docs should not be treated as a fixed list of every model you will see.
What happens after save
Once OpenAI is the system provider:
- normal chats can inherit it directly
- workspaces can keep using
System default - specific workspaces or agents can override it later
See Large Language Models for the override model.
Troubleshooting
The model list never loads
Check:
- the API key is valid
- the account has active billing or remaining quota
- the key has not been rotated or revoked
The key works elsewhere but RealTimeX shows no usable model
Some keys or organizations do not expose the same model catalog. Try reloading the page or selecting a different model that appears for that account.
You want to use an OpenAI-compatible gateway instead of OpenAI itself
Use Generic OpenAI instead of the native OpenAI provider page.