Azure config help

[XianYue0125]

What features would you like to see added?

The configuration is too complicated, with several configuration files. I need to use Microsoft's API with gpt-5-mini. I spent two hours configuring it, but it either doesn't have an Azure option or reports an error saying it can't find the model. I used dify and only needed to fill in two blanks to use it directly. I suggest optimizing this.

More details

Which components are impacted by your request?

No response

Pictures

No response

Code of Conduct

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

[danny-avila]

For azure, see this detailed configuration guide: https://www.librechat.ai/docs/configuration/azure

Make sure ENDPOINTS is configured to include azure as well, I.e. ENDPOINTS=azureOpenAI

I used dify and only needed to fill in two blanks to use it directly. I suggest optimizing this.

Thanks for the feedback. By nature, azure setup is not simple. The azure config was designed to allow for multiple deployments/regions to be used via the endpoint, especially before azure themselves started consolidating their endpoints into a single simple setup via the more “OpenAI-like” azure ai OpenAI endpoint. I will share some example configs that work and are pretty simple in a few minutes.

[danny-avila]

For example, in your librechat.yaml file, follow the steps here up to step 2: https://www.librechat.ai/docs/quick_start/custom_endpoints

Then add this (correspond to your azure deployment

endpoints:
  azureOpenAI:
    titleConvo: true
    groups:
    - group: "region-eastus-this-just-needs-to-be-unique" # arbitrary name
      apiKey: "${EASTUS_API_KEY}" # this maps to `EASTUS_API_KEY` set in .env file
      instanceName: "azure-eastus" # must match your actual region
      version: "2024-12-01-preview"
# assuming all these models are deployed, they should work
      models:
        gpt-4.1:
          deploymentName: "gpt-4.1"
        gpt-4.1-mini:
          deploymentName: "gpt-4.1-mini-eus2" # note, deployment name can differ from the openai model ID
        gpt-4.1-nano:
          deploymentName: "gpt-4.1-nano"
        gpt-5:
          deploymentName: "gpt-5"
        gpt-5-mini:
          deploymentName: "gpt-5-mini"
        gpt-5-nano:
          deploymentName: "gpt-5-nano"
        gpt-5-chat:
          deploymentName: "gpt-5-chat"

When visiting https://oai.azure.com, the instanceName should match what you see here:

Clicking "Deployments" on the left-hand side, you can see my corresponding deployments, where the 1st "name" column is deploymentName and the 2nd column "model name" matches the model key used in the config above:

[danny-avila]

In your .env file, you need to make sure these following values are defined.

ENDPOINTS=azureOpenAI # you can also add additional endpoints, like the default example:
# ENDPOINTS=openAI,agents,azureOpenAI,google,anthropic,custom
EASTUS_API_KEY=your-key-value

Note, you can rename EASTUS_API_KEY to whatever you want but they must match between .env and .yaml configs.

To get your key, click on any of the deployments and you can get it from this screen:

[xec-abailey]

Followed this closely and looks like I have everything setup accordingly. Thanks for the quick responses in this thread!

Still hitting connection errors when trying to get a response. Could it be the api version? I'm using 2025-01-01-preview

Edit: tried with 2024-12-01-preview also and it's still busted so must be something else... hmm

# For more information, see the Configuration Guide:
# https://www.librechat.ai/docs/configuration/librechat_yaml

# Configuration version (required)
version: 1.2.1

# Cache settings: Set to true to enable caching
cache: true

# File strategy s3/firebase
# fileStrategy: "s3"

# Custom interface configuration
interface:
  customWelcome: "Welcome to LibreChat! Enjoy your experience."
  # MCP Servers UI configuration

  fileSearch: true
  endpointsMenu: true
  modelSelect: true
  parameters: true
  sidePanel: true
  presets: true
  prompts: true
  bookmarks: true
  multiConvo: true
  agents: true
  temporaryChatRetention: 8

registration:
  socialLogins: ['github', 'saml']
  allowedDomains:
  - "coterra.com"

# Example MCP Servers Object Structure
  mcpServers:
    placeholder: 'MCP Servers'
  # Enable/disable file search as a chatarea selection (default: true)
  # Note: This setting does not disable the Agents File Search Capability.
  # To disable the Agents Capability, see the Agents Endpoint configuration instead.
#   everything:
#     # type: sse # type can optionally be omitted
#     url: http://localhost:3001/sse
#     timeout: 60000  # 1 minute timeout for this server, this is the default timeout for MCP servers.

# Definition of custom endpoints
  # assistants:
  #   disableBuilder: false # Disable Assistants Builder Interface by setting to `true`
  #   pollIntervalMs: 3000  # Polling interval for checking assistant updates
  #   timeoutMs: 180000  # Timeout for assistant operations
  #   # Should only be one or the other, either `supportedIds` or `excludedIds`
  #   supportedIds: ["asst_supportedAssistantId1", "asst_supportedAssistantId2"]
  #   # excludedIds: ["asst_excludedAssistantId"]
  #   # Only show assistants that the user created or that were created externally (e.g. in Assistants playground).
  #   # privateAssistants: false # Does not work with `supportedIds` or `excludedIds`
  #   # (optional) Models that support retrieval, will default to latest known OpenAI models that support the feature
  #   retrievalModels: ["gpt-4-turbo-preview"]
  #   # (optional) Assistant Capabilities available to all users. Omit the ones you wish to exclude. Defaults to list below.
  #   capabilities: ["code_interpreter", "retrieval", "actions", "tools", "image_vision"]
  # agents:
  #   # (optional) Default recursion depth for agents, defaults to 25
  #   recursionLimit: 50
  #   # (optional) Max recursion depth for agents, defaults to 25
  #   maxRecursionLimit: 100
  #   # (optional) Disable the builder interface for agents
  #   disableBuilder: false
  #   # (optional) Agent Capabilities available to all users. Omit the ones you wish to exclude. Defaults to list below.
  #   capabilities: ["execute_code", "file_search", "actions", "tools"]
endpoints:
  azureOpenAI:
    titleConvo: true
    titleModel: "gpt-4o-mini"
    summarize: true
    summaryModel: "gpt-4o-mini"
    plugins: true
    assistants: false

    groups:
      - group: "eastus2"
        apiKey: ${EASTUS_API_KEY}
        instanceName: ${INSTANCE_NAME}
        version: "2024-12-01-preview"
        models:
          gpt-5:
            deploymentName: "gpt-5"
            version: "2025-08-07"
          gpt-5-mini:
            deploymentName: "gpt-5-mini"
            version: "2025-08-07"
          gpt-5-nano:
            deploymentName: "gpt-5-nano"
            version: "2025-08-07"
          gpt-5-chat:
            deploymentName: "gpt-5-chat"
            version: "2025-08-07"
          gpt-4o:
            deploymentName: "gpt-4o"
            version: "2024-11-20"
          gpt-4o-mini:
            deploymentName: "gpt-4o-mini"
            version: "2024-07-18"
          gpt-4.1:
            deploymentName: "gpt-4.1"
            version: "2025-04-14"
          gpt-4.1-nano:
            deploymentName: "gpt-4.1-nano"
            version: "2025-04-14"
          gpt-4.1-mini:
            deploymentName: "gpt-4.1-mini"
            version: "2025-04-14"
          gpt-image-1:
            deploymentName: "gpt-image-1"
            version: "2025-04-15"
          o1:
            deploymentName: "o1"
            version: "2024-12-17"
          o3-mini:
            deploymentName: "o3-mini"
            version: "2025-01-31"
          o3:
            deploymentName: "o3"
            version: "2025-04-16"
          o4-mini:
            deploymentName: "o4-mini"
            version: "2025-04-16"
          gpt-4o-transcribe:
            deploymentName: "gpt-4o-transcribe"
            version: "2025-03-20"
          gpt-4o-mini-transcribe:
            deploymentName: "gpt-4o-mini-transcribe"
            version: "2025-03-20"
          gpt-4o-mini-tts:
            deploymentName: "gpt-4o-mini-tts"
            version: "2025-03-20"

# fileConfig:
#   endpoints:
#     assistants:
#       fileLimit: 5
#       fileSizeLimit: 10  # Maximum size for an individual file in MB
#       totalSizeLimit: 50  # Maximum total size for all files in a single request in MB
#       supportedMimeTypes:
#         - "image/.*"
#         - "application/pdf"
#     openAI:
#       disabled: true  # Disables file uploading to the OpenAI endpoint
#     default:
#       totalSizeLimit: 20
#     YourCustomEndpointName:
#       fileLimit: 2
#       fileSizeLimit: 5
#   serverFileSizeLimit: 100  # Global server file size limit in MB
#   avatarSizeLimit: 2  # Limit for user avatar image size in MB
#   imageGeneration: # Image Gen settings, either percentage or px
#     percentage: 100
#     px: 1024
#   # Client-side image resizing to prevent upload errors
#   clientImageResize:
#     enabled: false  # Enable/disable client-side image resizing (default: false)
#     maxWidth: 1900  # Maximum width for resized images (default: 1900)
#     maxHeight: 1900  # Maximum height for resized images (default: 1900)
#     quality: 0.92  # JPEG quality for compression (0.0-1.0, default: 0.92)
# # See the Custom Configuration Guide for more information on Assistants Config:
# # https://www.librechat.ai/docs/configuration/librechat_yaml/object_structure/assistants_endpoint

# Memory configuration for user memories
# memory:
#   # (optional) Disable memory functionality
#   disabled: false
#   # (optional) Restrict memory keys to specific values to limit memory storage and improve consistency
#   validKeys: ["preferences", "work_info", "personal_info", "skills", "interests", "context"]
#   # (optional) Maximum token limit for memory storage (not yet implemented for token counting)
#   tokenLimit: 10000
#   # (optional) Enable personalization features (defaults to true if memory is configured)
#   # When false, users will not see the Personalization tab in settings
#   personalize: true
#   # Memory agent configuration - either use an existing agent by ID or define inline
#   agent:
#     # Option 1: Use existing agent by ID
#     id: "your-memory-agent-id"
#     # Option 2: Define agent inline
#     # provider: "openai"
#     # model: "gpt-4o-mini"
#     # instructions: "You are a memory management assistant. Store and manage user information accurately."
#     # model_parameters:
#     #   temperature: 0.1

[marlonka]

2025-04-01-preview is the latest API version which works for us

[xec-abailey]

Did some additional testing:

1. Selected a model that does not use responses API to avoid any issue with that
2. tried using baseURL and serverless
3. added modelSpec for individual models that uses the azureOpenAI endpoint

Still no dice

[danny-avila]

You should share what you're seeing (screenshot) as well as any logs from the server.

Log locations:

• Docker: Project root directory ./logs
• npm: ./api/logs

There are two types of logs that can help diagnose the issue:

• debug logs (debug-YYYY-MM-DD.log)
• error logs (error-YYYY-MM-DD.log)

Error logs contain exact stack traces and are especially helpful, but both can provide valuable information.

Please only include the relevant portions of logs that correspond to when you reproduced the error.

[maxx-ukoo]

I also spent few hours and got 404 errror. I have
URL like https://*******************.cognitiveservices.azure.com/
api key
model gpt-40-mini
version

It is enough to run chat from python client but I can't configure LibreChat with these params.

[marlonka]

there is no model worldwide with your name "gpt-40-mini". either you mean "gpt-4o-mini" or just use gpt-4.1-mini instead

[maxx-ukoo]

yes, gpt-4o-mini correct parameter. But it doesn't work

[ismailuddin]

I too am having the same issue, I've followed all the instructions and ensured gotchas like the instance name match the exact resource name in Az AI Foundry, and making sure the right API version is selected, and the ENDPOINTS=azureOpenAI env variable is set.

My librechat.yaml contents:

endpoints:
  azureOpenAI:
    titleModel: 'gpt-4o-mini'
    groups:
      - group: 'azure'
        apiKey: "${AZURE_OPENAI_API_KEY}"
        instanceName: redacted
        version: "2025-01-01-preview"
        baseURL: "https://redacted.openai.azure.com/"
        models:
          gpt-4o-mini:
            deploymentName: gpt-4o-mini
            version: '2024-07-18'

These same details work in other tools e.g. Pydantic AI. I suspect either the docs are incorrect, or there is a bug with latest version of LibreChat, perhaps due to moving from env variables to the YAML for configuring Az OpenAI?

No logs in error log file, but debug log has following

2025-08-16T15:21:49.118Z info: RAG API is running and reachable at http://rag_api:8000.
2025-08-16T15:21:49.125Z info: Turnstile is DISABLED (no siteKey provided).
2025-08-16T15:21:53.365Z debug: [BaseClient] Loading history:
{
  conversationId: "0d0663eb-14dd-46a0-b292-791783a0fa54",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
}
2025-08-16T15:21:53.508Z debug: [BaseClient] Context Count (1/2)
{
  remainingContextTokens: 114742,
  maxContextTokens: 114750,
}
2025-08-16T15:21:53.508Z debug: [BaseClient] Context Count (2/2)
{
  remainingContextTokens: 114742,
  maxContextTokens: 114750,
}
2025-08-16T15:21:53.508Z debug: [BaseClient] tokenCountMap:
{
  9f64ccd5-026c-4ef9-8a6c-9ec7265419d0: 5,
}
2025-08-16T15:21:53.508Z debug: [BaseClient]
{
  promptTokens: 8,
  remainingContextTokens: 114742,
  payloadSize: 1,
  maxContextTokens: 114750,
}
2025-08-16T15:21:53.508Z debug: [BaseClient] tokenCountMap
{
  9f64ccd5-026c-4ef9-8a6c-9ec7265419d0: 5,
}
2025-08-16T15:21:53.509Z debug: [BaseClient] userMessage
{
  messageId: "9f64ccd5-026c-4ef9-8a6c-9ec7265419d0",
  parentMessageId: "00000000-0000-0000-0000-000000000000",
  conversationId: "0d0663eb-14dd-46a0-b292-791783a0fa54",
  sender: "User",
  text: "Hello",
  isCreatedByUser: true,
  tokenCount: 5,
}
2025-08-16T15:21:54.022Z debug: [spendTokens] conversationId: 0d0663eb-14dd-46a0-b292-791783a0fa54 | Context: message | Token usage:
{
  promptTokens: 8,
  completionTokens: 4,
}
2025-08-16T15:21:54.040Z debug: [spendTokens] No transactions incurred against balance
2025-08-16T15:22:01.688Z debug: [FinalizationRegistry] Cleaning up client for user 68925f7a04234604df4fd999

It would be nice if I could see exactly which API requests are being made, but not sure yet how to get that in debug logs...

[danny-avila]

    baseURL: "https://redacted.openai.azure.com/"

Don't use baseURL if you are using the default openai endpoint pattern (instance.openai.azure.com)

[ismailuddin]

I tried:

• Removing baseURL
• Putting the API key directly inside librechat.yaml
• Switching to API version 2025-04-01-preview as suggested above
• Double checked all names on AI Foundry...

but I still keep getting the exact same error... 😕 The log files have the same output as before. Inspecting the logs of the API container shows no errors (librechat.yaml is read successfully).

[danny-avila]

Please follow this as closely as possible, document how you followed the steps and how this led to your configuration.

• Azure config help #9009 (comment)

You can even share the docs with an AI to help confirm the configuration.
https://github.com/LibreChat-AI/librechat.ai/blob/main/pages/docs/configuration/azure.mdx

Also, share your logs at startup, not the ones from the chat request, to see if your configuration is being picked up correctly or if there are any warnings.

[ismailuddin]

Got it working! So the issue was even though on Az OpenAI the model version is specified as 2024-07-18, I cannot set this in librechat.yaml for a specific model.

Omitting baseURL is also key, if that is there the same error occurs.

endpoints:
  azureOpenAI:
    titleConvo: true
    groups:
      - group: 'azure'
        apiKey: "..."
        instanceName: redacted
        # version: "2024-12-01-preview"
        version: "2025-01-01-preview"
        models:
          gpt-4o-mini:
            deploymentName: gpt-4o-mini

Glad it's finally working! Thanks for your help! 😄

[Pfaufisch]

Maybe worth pointing it out explicitly, because I remember it confused me as well: The version in the config always refers to the Azure API version. The model version is not related to that and not relevant within the config.

https://www.librechat.ai/docs/configuration/azure#model-level-configuration

[ismailuddin]

Maybe worth pointing it out explicitly, because I remember it confused me as well: The version in the config always refers to the Azure API version. The model version is not related to that and not relevant within the config.

https://www.librechat.ai/docs/configuration/azure#model-level-configuration

Not so sure that's the issue. I was always setting the version field correctly at the group level using the API version. The real problem was specifying the model version, which despite being correct prevents the model from being discovered. Maybe it's more of an Azure problem...

[marlonka]

a newer API Version enables sometimes more features like thought summaries. please use 2025-04-01-preview for best support of recent models.

you select the model version when you deploy or edit the deployment in azure, so you can actually upgrade to a newer version of gpt-4o for example without restarting LibreChat

[coderReview]

Hello, I have been trying to use the azure configuration with a custom base url and it doesn't work. I've followed this example:

endpoints:
  azureOpenAI:
    titleModel: "gpt-4-turbo"
    groups:
      - group: "my-westus" # arbitrary name
        apiKey: "${WESTUS_API_KEY}"
        instanceName: "actual-instance-name" # name of the resource group or instance
        version: "2023-12-01-preview"
        # baseURL: https://prod.example.com
        # additionalHeaders:
        #   X-Custom-Header: value
        models:
          gpt-4-vision-preview:
            deploymentName: gpt-4-vision-preview
            version: "2024-02-15-preview"
          gpt-3.5-turbo:
            deploymentName: gpt-35-turbo
          gpt-3.5-turbo-1106:
            deploymentName: gpt-35-turbo-1106
          gpt-4:
            deploymentName: gpt-4
          gpt-4-1106-preview:
            deploymentName: gpt-4-1106-preview
      - group: "my-eastus"
        apiKey: "${EASTUS_API_KEY}"
        instanceName: "actual-eastus-instance-name"
        deploymentName: gpt-4-turbo
        version: "2024-02-15-preview"
        baseURL: "https://gateway.ai.cloudflare.com/v1/cloudflareId/azure/azure-openai/${INSTANCE_NAME}/${DEPLOYMENT_NAME}" # uses env variables
        additionalHeaders:
          X-Custom-Header: value
        models:
          gpt-4-turbo: true

The issue seems to be in the extractBaseURL method in extractBaseURL.js file. The baseURL ignores everything after "azure-openai". So the instance name and deployment name variables are ignored.

Right now, I modified the extractBaseURL method to return the baseURL as it was provided in the YML file.

Any ideas? Thank you