Unable to get titles for o4-mini-deep-research #9506
-
What happened?When I run deep research, chats end up with the title "New Chat". I see the error: Reading through the discussion here, I see that most likely the "Use Responses API" setting is the culprit. In the above example, it is My .env file has this relevant data: Version Informationdocker images | grep librechat Steps to Reproduce
What browsers are you seeing the problem on?No response Relevant log output2025-09-07 08:24:07 warn: [api/server/controllers/agents/client.js #titleConvo] Error getting endpoint config
2025-09-07 08:24:08 error: [api/server/controllers/agents/client.js #titleConvo] Error 404 This model is only supported in v1/responses and not in v1/chat/completions.
Troubleshooting URL: https://js.langchain.com/docs/troubleshooting/errors/MODEL_NOT_FOUND/
2025-09-07 08:29:38 error: [api/server/controllers/agents/client.js #sendCompletion] Operation aborted Aborted
2025-09-07 09:08:58 error: [api/server/controllers/agents/client.js #sendCompletion] Operation aborted Aborted
2025-09-07 09:15:19 warn: [api/server/controllers/agents/client.js #titleConvo] Error getting endpoint config
2025-09-07 09:15:20 error: [api/server/controllers/agents/client.js #titleConvo] Error 404 This model is only supported in v1/responses and not in v1/chat/completions.
Troubleshooting URL: https://js.langchain.com/docs/troubleshooting/errors/MODEL_NOT_FOUND/
2025-09-07 09:23:13 warn: [api/server/controllers/agents/client.js #titleConvo] Error getting endpoint config
2025-09-07 09:23:14 error: [api/server/controllers/agents/client.js #titleConvo] Error 404 This model is only supported in v1/responses and not in v1/chat/completions.
Troubleshooting URL: https://js.langchain.com/docs/troubleshooting/errors/MODEL_NOT_FOUND/Screenshots
Code of Conduct
|
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
By default, LibreChat will attempt to both use the current model and chat/completions endpoint for structured outputs based on the conversation to generate a title. To customize both of these, including the system prompt used for titling, you can use these settings: It is highly recommended to use a smaller/faster model than deep-research models for titling, as they can also timeout as title generations are cancelled if they take too long. |
Beta Was this translation helpful? Give feedback.
-
|
That worked! Thanks a million! |
Beta Was this translation helpful? Give feedback.
By default, LibreChat will attempt to both use the current model and chat/completions endpoint for structured outputs based on the conversation to generate a title.
To customize both of these, including the system prompt used for titling, you can use these settings:
https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/shared_endpoint_settings#titlemodel
It is highly recommended to use a smaller/faster model than deep-research models for titling, as they can also timeout as title generations are cancelled if they take too long.