Azure OpenAI: Invalid parameter 'response_format' of type 'json_schema' appears after updating to 0.4.0

[KoreanThinker]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain.js documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain.js rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

import * as hub from "langchain/hub";
import azureOpenai from "@/config/azureOpenai";

export default async function personalisationEvaluation(
  userProfile,
  question,
  answer,
) {
  const prompt = await hub.pull("evaluate-personalisation");
  const model = prompt.pipe(azureOpenai);

  const result = await model.invoke(
    {
      user_profile: userProfile,
      user_question: question,
      model_answer: answer,
    },
    { runName: "EvaluatePersonalisation" },
  );

  return {
    personalisationScore: result.score,
    personalisationExplanation: result.explanation,
  };
}

Error Message and Stack Trace (if applicable)

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
at oL.generate (.next/server/app/api/chat-bot/callback/route.js:400:30952)
at uY.makeStatusError (.next/server/app/api/chat-bot/callback/route.js:400:44061)
at uY.makeRequest (.next/server/app/api/chat-bot/callback/route.js:400:45005)
at async (.next/server/app/api/chat-bot/callback/route.js:411:17970)
at async t._fn (.next/server/app/api/chat-bot/callback/route.js:912:254750) {
status: 400,
headers: [Object],
request_id: '3d372b97-b1a9-48bb-93cb-5346235e4f80',
error: [Object],
code: null,
param: null,
type: 'invalid_request_error',
attemptNumber: 1,
retriesLeft: 6
}

Description

When I use version 0.3.17, everything works perfectly without any issues. However, after updating to 0.4.0, I consistently encounter the error: "Invalid parameter: 'response_format' of type 'json_schema'".

It seems likely that this issue is related to changes introduced in #7535.

Interestingly, prompts using the same model with only slight differences in output structure work without any problems. This inconsistency makes the issue even more puzzling.

System Info

node = v22.11.0
yarn = v1.22.22
"@langchain/community": "^0.3.27",
"@langchain/core": "^0.3.35",
"@langchain/google-genai": "^0.1.6",
"@langchain/openai": "^0.4.0",
"@langchain/pinecone": "^0.1.3",
"langchain": "^0.3.13",
"langsmith": "^0.3.3",

[dosubot]

The error "Invalid parameter: 'response_format' of type 'json_schema'" occurs because, in LangChain.js version 0.4.0, the responseFormat parameter is expected to be either "content" or "content_and_artifact" ^[1]. The use of "json_schema" is no longer supported in this version, which is why you're encountering this error after updating.

To resolve this issue, you should update your code to use one of the supported responseFormat values. If your application relies on structured outputs, you may need to adjust your implementation to align with the new response format handling in version 0.4.0.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[jacoblee93]

Hey @KoreanThinker, thanks for reporting - we switched the default method for structured output to jsonSchema, but it seems some of the older Azure deployments don't support it yet. Will revert it for Azure for now.

[jacoblee93]

And just confirming - you're using gpt-4o as your model right? I see it in your screenshot

[jacoblee93]

Just shipped 0.4.1. Let me know if that fixes it for you!

[KoreanThinker]

Thank you for resolving the issue! Your quick response was incredibly helpful. Looking forward to more great updates in the future.