Earlier, we were using the GPT-4o model and the agent was running properly. Currently, the model has been changed to GPT-5.4, and the agent is not running as it is showing a timeout error. We are also unable to change the model back to GPT-4o, as there is no option available to do so. Please let help me with the quick solution.
Hi @snehamayi.senapati ,
You can try increasing the agent/request timeout duration, if configurable, as GPT-5.4 responses may take longer compared to GPT-4o.
Thanks,
Could you please help how to do that .it will help me to run the agent
Hi,
The timeout may vary based on the below points. Have you tried these options?
1] Reduce the system prompt size
→ Try keeping only the essential instructions in the prompt.
2] Reduce tool call complexity
→ If the agent is invoking multiple tools/workflows, try splitting them into smaller steps.
Thanks,
I am not sure why you are unable to change the model back 4.o
Click on that model dropdown and scroll down there you can 4.o models you can select required one.
Happy Automation.
Not i am not able to switch back to 4.o no option is showing? could anyone help me to change back the model
@snehamayi.senapati what are the options that you can see? try any lower model or higher if available in the dropdown. can you send the screenshot that whatever options you can see
From the screenshot, it looks like the model dropdown is not loading properly or the tenant/model access is restricted for your current setup.
You can try the following:
- Refresh the browser and reopen Orchestrator
- Clear browser cache or try Incognito mode
- Verify whether the required AI/Agent models are enabled for your tenant
- Check if your organization admin has restricted model access/policies
- Try opening the Agent configuration from another browser/device
If no models appear in the dropdown at all, this is likely:
- a temporary platform issue,
- permission restriction,
- or tenant configuration issue rather than a workflow problem.
You may need to contact your tenant admin or UiPath support to re-enable/access the available models.
Hi @snehamayi.senapati ,
I am also unable to find any other models. It seems like these models are retiring from OpenAI. Please refer this link and the below image.
However, you can still try to access them using Open AI connection. But you need to subscribe for Tier1 or above in OpenAI. Then you need to create OpenAI connection in Orchestrator.
-
Use Custom Model
-
Select Connection
-
Select and Test the Model
Thanks for the response.. This needs Open AI API key. i dont have open AI account so how can i implement back to 4o version
You need to sign in to this OpenAI api portal link with your Google account. Then you will be able to generate api key. However you need to subscribe to tier 1 or above to access gpt 4o.
The issue is happening because some older OpenAI models are being retired, so they may no longer appear directly in UiPath. To continue using GPT-4o, you need to create an OpenAI API account, generate an API key, and configure it through an OpenAI connection in Orchestrator. After that, you can select the custom model and test the connection from UiPath.
Also, GPT-4o access is generally available only for OpenAI accounts with Tier 1 or higher usage access, so without that subscription the model may not be accessible.