Agent does not create Convo Title

[peeeteeer]

What happened?

since recently an Agent does not create a conversation title... just creates "New Chat" - it works fine when using the models directly in a chat, it fails only in Agent.

Version Information

150116e

Steps to Reproduce

1.) create an Agent
2.) start a conversation

What browsers are you seeing the problem on?

No response

Relevant log output

2025-04-26T15:56:49.605Z error: [api/server/controllers/agents/client.js #titleConvo] Error 400 Invalid parameter: 'response_format' of type 'json_schema' is not supported with this ... [truncated]

{"attemptNumber":1,"code":null,"error":{"code":null,"message":"Invalid parameter: 'response_format' of type 'json_schema' is not supported with this model. Learn more about supported models at the Structured Outputs guide: https://platform.openai.com/docs/guides/structured-outputs","param":null,"type":"invalid_request_error"},"headers":{"apim-request-id":"144e72ed-a266-4974-be15-dd20ac9237bd","azureml-model-session":"v20250415-5-168444713-4","content-length":"330","content-type":"application/json","date":"Sat, 26 Apr 2025 15:56:49 GMT","ms-azureml-model-error-reason":"model_error","ms-azureml-model-error-statuscode":"400","server":"Kestrel","strict-transport-security":"max-age=31536000; includeSubDomains; preload","x-content-type-options":"nosniff","x-ms-client-request-id":"144e72ed-a266-4974-be15-dd20ac9237bd","x-ms-deployment-name":"PTU_deployment","x-ms-rai-invoked":"true","x-ms-region":"Sweden Central","x-request-id":"ee773f46-0f39-4b1b-96c3-76d8a91cc4b8"},"level":"error","message":"[api/server/controllers/agents/client.js #titleConvo] Error 400 Invalid parameter: 'response_format' of type 'json_schema' is not supported with this model. Learn more about supported models at the Structured Outputs guide: https://platform.openai.com/docs/guides/structured-outputs","param":null,"request_id":"ee773f46-0f39-4b1b-96c3-76d8a91cc4b8","retriesLeft":6,"stack":"Error: 400 Invalid parameter: 'response_format' of type 'json_schema' is not supported with this model. Learn more about supported models at the Structured Outputs guide: https://platform.openai.com/docs/guides/structured-outputs\n    at APIError.generate (/app/node_modules/openai/error.js:45:20)\n    at CustomAzureOpenAIClient.makeStatusError (/app/node_modules/openai/core.js:302:33)\n    at CustomAzureOpenAIClient.makeRequest (/app/node_modules/openai/core.js:346:30)\n    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n    at async /app/node_modules/@librechat/agents/node_modules/@langchain/openai/dist/chat_models.cjs:2036:29\n    at async RetryOperation._fn (/app/node_modules/p-retry/index.js:50:12)","status":400,"type":"invalid_request_error"}

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[danny-avila]

What model are you using? you should switch the title model as needed. No issues here

[peeeteeer]

titleModel: "gpt-4o-mini" works     titleModel: "gpt-4o" does not work - but only for Agents... in a normal Chat everything is fine...

[dartungar]

Same here.
I am using o4-mini agent with Google tool enabled.
Here's what I get in the server-side logs:

2025-04-28T10:50:51.672202190Z 2025-04-28 10:50:51 error: Title error: Title generation timeout
2025-04-28T10:50:51.674469361Z 2025-04-28 10:50:51 error: [api/server/controllers/agents/client.js #titleConvo] Error Aborted

...which does not give much information.

when I change agent's model to Claude Sonnet 3.7 it works fine, though.

[joshuamckay]

Same here for agents using openai models with Google tool enabled.

[0xvm]

I think this occurs due to OpenAI reasoning models (o1, o3, o4) not supporting temperatures. Yet, in the titleConvo method the title-generation payload is built with:

const modelOptions = {
  model,
  temperature: 0.2,
  presence_penalty: 0,
  frequency_penalty: 0,
  max_tokens: 16,
};

https://github.com/danny-avila/LibreChat/blob/main/api/app/clients/OpenAIClient.js#L743

if we can set it to 1.0 (or unset it altogether so the default is used?) that might do the trick?

debug.log snippet:

2025-06-12T23:34:10.368Z error: [RunManager] handleLLMError: {"context":"title","tokenBuffer":150,"conversationId":"d9845a4c-e0b7-4f85-8efe-a560600f8dbd"} 400 Unsupported value: 'tem... [truncated]
2025-06-12T23:34:10.369Z error: [OpenAIClient] There was an issue generating title with LangChain, trying completion method... 400 Unsupported value: 'temperature' does not support 0... [truncated]
2025-06-12T23:34:10.372Z debug: [OpenAIClient] chatCompletion
{
  baseURL: "https://api.openai.com/v1",
    modelOptions.model: "o4-mini",
    modelOptions.user: "6846b9b79e0b0a3e620ca8db",
    modelOptions.temperature: 0.2,
    modelOptions.presence_penalty: 0,
    modelOptions.frequency_penalty: 0,
    modelOptions.max_tokens: 16,
    // 1 message(s)
    modelOptions.messages: [{"role":"system","content":"Please generate a concise, 5-word-or-less title for the conversation, us... [truncated]],
}

error.log snippet:

{"attemptNumber":1,"code":"unsupported_value","error":{"code":"unsupported_value","message":"Unsupported value: 'temperature' does not support 0.2 with this model. Only the default (1) value is supported.","param":"temperature","type":"invalid_request_error"},

[vlad-tsoy]

With the introduction of the Responses API, I noticed that the agents that use reasoning models have a conversation title broken, as well as a regular chat with a reasoning model. We exclusively use Azure OpenAI, and my logs show this:

{"level":"error","message":"[api/server/controllers/agents/client.js #titleConvo] Error The OPENAI_API_KEY environment variable is missing or empty; either provide it, or instantiate the OpenAI client with an apiKey option, like new OpenAI({ apiKey: 'My API Key' }).","stack":"Error: The OPENAI_API_KEY environment variable is missing or empty; either provide it, or instantiate the OpenAI client with an apiKey option, like new OpenAI({ apiKey: 'My API Key' }).\n at new OpenAI (/app/node_modules/@langchain/openai/node_modules/openai/client.js:86:19)\n at new CustomOpenAIClient (/app/node_modules/@librechat/agents/dist/cjs/llm/openai/index.cjs:53:1)\n at ChatOpenAI._getClientOptions (/app/node_modules/@librechat/agents/dist/cjs/llm/openai/index.cjs:126:27)\n at ChatOpenAI.betaParsedCompletionWithRetry (/app/node_modules/@langchain/openai/dist/chat_models.cjs:2443:37)\n at ChatOpenAI._generate (/app/node_modules/@langchain/openai/dist/chat_models.cjs:2226:35)\n at /app/node_modules/@langchain/core/dist/language_models/chat_models.cjs:254:96\n at Array.map ()\n at ChatOpenAI._generateUncached (/app/node_modules/@langchain/core/dist/language_models/chat_models.cjs:254:67)\n at process.processTicksAndRejections (node:internal/process/task_queues:95:5)\n at async ChatOpenAI.invoke (/app/node_modules/@langchain/core/dist/language_models/chat_models.cjs:92:24)"}

[andreimindru]

I have a similar issue! It tries to use OpenAI instead of Azure OpenAI for title generation when using an agent with an Azure OpenAI model behind it

[vlad-tsoy]

pull the latest image the issue was resolved.

[andreimindru]

Thank you for the incredibly fast reply!