Azure OpenAI Chat Model: Add Header

Todd_Brown · October 15, 2025, 3:30pm

Is there a way to add a customer header to the Azure OpenAI Chat Model? I am using an “internal” version of OpenAI which requires a specific header on the API requests to track usage. Otherwise it fails as “BadRequest”.

Todd_Brown · October 16, 2025, 6:59pm

I think the code in question is located here: n8n/packages/@n8n/nodes-langchain/nodes/llms/LmChatAzureOpenAi at b21140180e32c335ad0392adf618d33ba05e2838 · n8n-io/n8n · GitHub

and it constructs the model proxy (AzureChatOpenAI)

// Create and return the model
const model = new AzureChatOpenAI({
    azureOpenAIApiDeploymentName: modelName,
    ...modelConfig,
    ...options,
    timeout: options.timeout ?? 60000,
    maxRetries: options.maxRetries ?? 2,
    callbacks: [new N8nLlmTracing(this)],
    configuration: {
        fetchOptions: {
            dispatcher: getProxyAgent(),
        },
    },
    modelKwargs: options.responseFormat
        ? {
                response_format: { type: options.responseFormat },
            }
        : undefined,
    onFailedAttempt: makeN8nLlmFailedAttemptHandler(this),
});

So I think that is this class: langchainjs/libs/providers/langchain-openai/src/azure/chat_models.ts at 1d95681731d5c09682350b84f6cbf1b6906f6761 · langchain-ai/langchainjs · GitHub py which looks like it supports headers

    const defaultHeaders = normalizeHeaders(params.defaultHeaders);
    params.defaultHeaders = {
      ...params.defaultHeaders,
      "User-Agent": defaultHeaders["User-Agent"]
        ? `langchainjs-azure-openai/2.0.0 (${env})${defaultHeaders["User-Agent"]}`
        : `langchainjs-azure-openai/2.0.0 (${env})`,
    };

Linking the documentation which does discuss custom headers:

const llmWithCustomHeaders = new AzureChatOpenAI({
  azureOpenAIApiKey: process.env.AZURE_OPENAI_API_KEY, // In Node.js defaults to process.env.AZURE_OPENAI_API_KEY
  azureOpenAIApiInstanceName: process.env.AZURE_OPENAI_API_INSTANCE_NAME, // In Node.js defaults to process.env.AZURE_OPENAI_API_INSTANCE_NAME
  azureOpenAIApiDeploymentName: process.env.AZURE_OPENAI_API_DEPLOYMENT_NAME, // In Node.js defaults to process.env.AZURE_OPENAI_API_DEPLOYMENT_NAME
  azureOpenAIApiVersion: process.env.AZURE_OPENAI_API_VERSION, // In Node.js defaults to process.env.AZURE_OPENAI_API_VERSION
  configuration: {
    defaultHeaders: {
      "x-custom-header": `SOME_VALUE`,
    },
  },
});

Todd_Brown · October 16, 2025, 7:34pm

I am guessing you could take the header bits from n8n/packages/nodes-base/nodes/HttpRequest/V3/Description.ts:

and paste them LmChatAzureOpenAi/properties: n8n/packages/@n8n/nodes-langchain/nodes/llms/LmChatAzureOpenAi/properties.ts:

github.com/n8n-io/n8n

packages/%40n8n/nodes-langchain/nodes/llms/LmChatAzureOpenAi/properties.ts

b21140180

import type { INodeProperties } from 'n8n-workflow';
import { NodeConnectionTypes } from 'n8n-workflow';

import { getConnectionHintNoticeField } from '@utils/sharedFields';

import { AuthenticationType } from './types';

export const properties: INodeProperties[] = [
	// eslint-disable-next-line n8n-nodes-base/node-param-default-missing
	{
		displayName: 'Authentication',
		name: 'authentication',
		type: 'options',
		default: AuthenticationType.ApiKey,
		options: [
			{
				name: 'API Key',
				value: AuthenticationType.ApiKey,
			},
			{

This file has been truncated. show original

and refernce them from LmChatAzureOpenAi:

github.com/n8n-io/n8n

packages/%40n8n/nodes-langchain/nodes/llms/LmChatAzureOpenAi/LmChatAzureOpenAi.node.ts

b21140180

import { AzureChatOpenAI } from '@langchain/openai';
import {
	NodeOperationError,
	NodeConnectionTypes,
	type INodeType,
	type INodeTypeDescription,
	type ISupplyDataFunctions,
	type SupplyData,
} from 'n8n-workflow';

import { getProxyAgent } from '@utils/httpProxyAgent';

import { setupApiKeyAuthentication } from './credentials/api-key';
import { setupOAuth2Authentication } from './credentials/oauth2';
import { properties } from './properties';
import { AuthenticationType } from './types';
import type {
	AzureOpenAIApiKeyModelConfig,
	AzureOpenAIOAuth2ModelConfig,
	AzureOpenAIOptions,

This file has been truncated. show original

but I really don’t know - just guessing at this a bit. Would love some guidance here.

Todd_Brown · October 18, 2025, 12:01pm

So i installed 1.112.0

npm install -g [email protected]

because I was pointed towards this feat(OpenAI Node): Support custom headers for model requests by mgosal · Pull Request #17835 · n8n-io/n8n · GitHub

feat(OpenAI Node): Support custom headers for model requests
github-actions bot mentioned this pull request on Sep 15

Release 1.112.0 #19556

And confirmed it:

I guess maybe I am confusing AzureOpenAi and OpenAI in this ecosystem… maybe? @mgosal @idirouhab any chance you can weigh in here?

eugene.molodkin · January 14, 2026, 7:46pm

Hey, Todd!

Unfortunately custom header is only supported by OpenAI Chat Model node. You can still use this node for OpenAI-compatible API providers by overriding base URL in OpenAI credentials.

But as I understand your internal deployment is on Azure, and Azure connection is a bit different than regular OpenAI-compatible APIs, thus requires using Azure OpenAI Chat Model (and underlying AzureChatOpenAI class from langchain).

If you use self-hosted n8n – you can try using LangChain code node as a workaround. This node allows you to write custom implementations for AI nodes, including models.

Here’s a brief example to get started: