Azure OpenAI Service REST API reference

Article
07/09/2024

This article provides details on the inference REST API endpoints for Azure OpenAI.

API specs

Managing and interacting with Azure OpenAI models and resources is divided across three primary API surfaces:

Control plane
Data plane - authoring
Data plane - inference

Each API surface/specification encapsulates a different set of Azure OpenAI capabilities. Each API has its own unique set of preview and stable/generally available (GA) API releases. Preview releases currently tend to follow a monthly cadence.

API	Latest preview release	Latest GA release	Specifications	Description
Control plane	`2024-04-01-preview`	`2023-05-01`	Spec files	Azure OpenAI shares a common control plane with all other Azure AI Services. The control plane API is used for things like creating Azure OpenAI resources, model deployment, and other higher level resource management tasks. The control plane also governs what is possible to do with capabilities like Azure Resource Manager, Bicep, Terraform, and Azure CLI.
Data plane - authoring	`2024-05-01-preview`	`2024-06-01`	Spec files	The data plane authoring API controls fine-tuning, file-upload, ingestion jobs, and certain model level queries
Data plane - inference	`2024-05-01-preview`	`2024-06-01`	Spec files	The data plane inference API provides the inference capabilities/endpoints for features like completions, chat completions, embeddings, speech/whisper, on your data, Dall-e, assistants, etc.

Authentication

Azure OpenAI provides two methods for authentication. You can use either API Keys or Microsoft Entra ID.

API Key authentication: For this type of authentication, all API requests must include the API Key in the api-key HTTP header. The Quickstart provides guidance for how to make calls with this type of authentication.
Microsoft Entra ID authentication: You can authenticate an API call using a Microsoft Entra token. Authentication tokens are included in a request as the Authorization header. The token provided must be preceded by Bearer, for example Bearer YOUR_AUTH_TOKEN. You can read our how-to guide on authenticating with Microsoft Entra ID.

REST API versioning

The service APIs are versioned using the api-version query parameter. All versions follow the YYYY-MM-DD date structure. For example:

POST https://YOUR_RESOURCE_NAME.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME/completions?api-version=2024-06-01

Data plane inference

The rest of the article covers the latest GA release of the Azure OpenAI data plane inference specification, 2024-06-01.

If you're looking for documentation on the latest preview API release, refer to the latest preview data plane inference API

Completions

POST https://{endpoint}/openai/deployments/{deployment-id}/completions?api-version=2024-06-01

Creates a completion for the provided prompt, parameters and chosen model.

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string	Deployment id of the model which was deployed.
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
prompt	string or array	The prompt(s) to generate completions for, encoded as a string or array of strings. Note that <\|endoftext\|> is the document separator that the model sees during training, so if a prompt isn'tspecified the model will generate as if from the beginning of a new document. Maximum allowed size of string list is 2048.	No
max_tokens	integer	The token count of your prompt plus max_tokens can't exceed the model's context length. Most models have a context length of 2048 tokens (except for the newest models, which support 4096). Has minimum of 0.	No	16
temperature	number	What sampling temperature to use. Higher values means the model will take more risks. Try 0.9 for more creative applications, and 0 (arg max sampling) for ones with a well-defined answer. We generally recommend altering this or top_p but not both.	No	1
top_p	number	An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or temperature but not both.	No	1
logit_bias	object	Defaults to null. Modify the likelihood of specified tokens appearing in the completion. Accepts a json object that maps tokens (specified by their token ID in the GPT tokenizer) to an associated bias value from -100 to 100. You can use this tokenizer tool (which works for both GPT-2 and GPT-3) to convert text to token IDs. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token. As an example, you can pass {"50256":-100} to prevent the <\|endoftext\|> token from being generated.	No
user	string	A unique identifier representing your end-user, which can help monitoring and detecting abuse	No
n	integer	How many completions to generate for each prompt. Minimum of 1 and maximum of 128 allowed. Note: Because this parameter generates many completions, it can quickly consume your token quota. Use carefully and ensure that you have reasonable settings for max_tokens and stop.	No	1
stream	boolean	Whether to stream back partial progress. If set, tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a data: [DONE] message.	No	False
logprobs	integer	Include the log probabilities on the logprobs most likely tokens, as well the chosen tokens. For example, if logprobs is 5, the API will return a list of the 5 most likely tokens. The API will always return the logprob of the sampled token, so there may be up to logprobs+1 elements in the response. Minimum of 0 and maximum of 5 allowed.	No	None
suffix	string	The suffix that comes after a completion of inserted text.	No
echo	boolean	Echo back the prompt in addition to the completion	No	False
stop	string or array	Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence.	No
completion_config	string		No
presence_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.	No	0
frequency_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.	No	0
best_of	integer	Generates best_of completions server-side and returns the "best" (defined as the one with the highest log probability per token). Results can't be streamed. When used with n, best_of controls the number of candidate completions and n specifies how many to return - best_of must be greater than n. Note: Because this parameter generates many completions, it can quickly consume your token quota. Use carefully and ensure that you have reasonable settings for max_tokens and stop. Has maximum value of 128.	No

Responses

Name	Type	Description	Required
id	string		Yes
object	string		Yes
created	integer		Yes
model	string		Yes
prompt_filter_results	promptFilterResults	Content filtering results for zero or more prompts in the request. In a streaming request, results for different prompts may arrive at different times or in different orders.	No
choices	array		Yes
usage	object		No

Properties for usage

completion_tokens

Name	Type	Description	Default
completion_tokens	number
prompt_tokens	number
total_tokens	number

Status Code: 200

Description: OK

Content-Type	Type	Description
application/json	object

Status Code: default

Description: Service unavailable

Content-Type	Type	Description
application/json	errorResponse

Examples

Example

Creates a completion for the provided prompt, parameters and chosen model.

POST https://{endpoint}/openai/deployments/{deployment-id}/completions?api-version=2024-06-01

{
 "prompt": [
  "tell me a joke about mango"
 ],
 "max_tokens": 32,
 "temperature": 1.0,
 "n": 1
}

Responses: Status Code: 200

{
  "body": {
    "id": "cmpl-7QmVI15qgYVllxK0FtxVGG6ywfzaq",
    "created": 1686617332,
    "choices": [
      {
        "text": "es\n\nWhat do you call a mango who's in charge?\n\nThe head mango.",
        "index": 0,
        "finish_reason": "stop",
        "logprobs": null
      }
    ],
    "usage": {
      "completion_tokens": 20,
      "prompt_tokens": 6,
      "total_tokens": 26
    }
  }
}

Embeddings

POST https://{endpoint}/openai/deployments/{deployment-id}/embeddings?api-version=2024-06-01

Get a vector representation of a given input that can be easily consumed by machine learning models and algorithms.

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: application/json

Name	Type	Description	Required
input	string or array	Input text to get embeddings for, encoded as a string. To get embeddings for multiple inputs in a single request, pass an array of strings. Each input must not exceed 2048 tokens in length. Unless you're embedding code, we suggest replacing newlines (\n) in your input with a single space, as we have observed inferior results when newlines are present.	Yes
user	string	A unique identifier representing your end-user, which can help monitoring and detecting abuse.	No
input_type	string	input type of embedding search to use	No
encoding_format	string	The format to return the embeddings in. Can be either `float` or `base64`. Defaults to `float`.	No
dimensions	integer	The number of dimensions the resulting output embeddings should have. Only supported in `text-embedding-3` and later models.	No

Responses

Name	Type	Required
object	string	Yes
model	string	Yes
data	array	Yes
usage	object	Yes

Properties for usage

prompt_tokens

Name	Type	Description	Default
prompt_tokens	integer

total_tokens

Name	Type	Description	Default
total_tokens	integer

Status Code: 200

Description: OK

Content-Type	Type	Description
application/json	object

Examples

Example

Return the embeddings for a given prompt.

POST https://{endpoint}/openai/deployments/{deployment-id}/embeddings?api-version=2024-06-01

{
 "input": [
  "this is a test"
 ]
}

Responses: Status Code: 200

{
  "body": {
    "data": [
      {
        "index": 0,
        "embedding": [
          -0.012838088,
          -0.007421397,
          -0.017617522,
          -0.028278312,
          -0.018666342,
          0.01737855,
          -0.01821495,
          -0.006950092,
          -0.009937238,
          -0.038580645,
          0.010674067,
          0.02412286,
          -0.013647936,
          0.013189907,
          0.0021125758,
          0.012406612,
          0.020790534,
          0.00074595667,
          0.008397198,
        ]
      }
    ],
    "usage": {
      "prompt_tokens": 4,
      "total_tokens": 4
    }
  }
}

Chat completions

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-06-01

Creates a completion for the chat message

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string	Deployment id of the model which was deployed.
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
temperature	number	What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. We generally recommend altering this or `top_p` but not both.	No	1
top_p	number	An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or `temperature` but not both.	No	1
stream	boolean	If set, partial message deltas will be sent, like in ChatGPT. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a `data: [DONE]` message.	No	False
stop	string or array	Up to 4 sequences where the API will stop generating further tokens.	No
max_tokens	integer	The maximum number of tokens allowed for the generated answer. By default, the number of tokens the model can return will be (4096 - prompt tokens).	No	4096
presence_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.	No	0
frequency_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.	No	0
logit_bias	object	Modify the likelihood of specified tokens appearing in the completion. Accepts a json object that maps tokens (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token.	No
user	string	A unique identifier representing your end-user, which can help Azure OpenAI to monitor and detect abuse.	No
messages	array	A list of messages comprising the conversation so far. Example Python code.	No
data_sources	array	The configuration entries for Azure OpenAI chat extensions that use them. This additional specification is only compatible with Azure OpenAI.	No
n	integer	How many chat completion choices to generate for each input message.	No	1
seed	integer	If specified, our system will make a best effort to sample deterministically, such that repeated requests with the same `seed` and parameters should return the same result. Determinism isn'tguaranteed, and you should refer to the `system_fingerprint` response parameter to monitor changes in the backend.	No	0
logprobs	boolean	Whether to return log probabilities of the output tokens or not. If true, returns the log probabilities of each output token returned in the `content` of `message`. This option is currently not available on the `gpt-4-vision-preview` model.	No	False
top_logprobs	integer	An integer between 0 and 5 specifying the number of most likely tokens to return at each token position, each with an associated log probability. `logprobs` must be set to `true` if this parameter is used.	No
response_format	object	An object specifying the format that the model must output. Used to enable JSON mode.	No
tools	array	A list of tools the model may call. Currently, only functions are supported as a tool. Use this to provide a list of functions the model may generate JSON inputs for.	No
tool_choice	chatCompletionToolChoiceOption	Controls which (if any) function is called by the model. `none` means the model will not call a function and instead generates a message. `auto` means the model can pick between generating a message or calling a function. Specifying a particular function via `{"type": "function", "function": {"name": "my_function"}}` forces the model to call that function.	No
functions	array	Deprecated in favor of `tools`. A list of functions the model may generate JSON inputs for.	No
function_call	string or object	Deprecated in favor of `tool_choice`. Controls how the model responds to function calls. "none" means the model doesn't call a function, and responds to the end-user. "auto" means the model can pick between an end-user or calling a function. Specifying a particular function via `{"name":\ "my_function"}` forces the model to call that function. "none" is the default when no functions are present. "auto" is the default if functions are present.	No

Properties for response_format

Type

Name	Type	Description	Default
type	chatCompletionResponseFormat	Setting to `json_object` enables JSON mode. This guarantees that the message the model generates is valid JSON.	text

Responses

Status Code: 200

Description: OK

Content-Type	Type	Description
application/json	createChatCompletionResponse

Status Code: default

Description: Service unavailable

Content-Type	Type	Description
application/json	errorResponse

Examples

Example

Creates a completion for the provided prompt, parameters and chosen model.

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-06-01

{
 "messages": [
  {
   "role": "system",
   "content": "you're a helpful assistant that talks like a pirate"
  },
  {
   "role": "user",
   "content": "can you tell me how to care for a parrot?"
  }
 ]
}

Responses: Status Code: 200

{
  "body": {
    "id": "chatcmpl-7R1nGnsXO8n4oi9UPz2f3UHdgAYMn",
    "created": 1686676106,
    "choices": [
      {
        "index": 0,
        "finish_reason": "stop",
        "message": {
          "role": "assistant",
          "content": "Ahoy matey! So ye be wantin' to care for a fine squawkin' parrot, eh? Well, shiver me timbers, let ol' Cap'n Assistant share some wisdom with ye! Here be the steps to keepin' yer parrot happy 'n healthy:\n\n1. Secure a sturdy cage: Yer parrot be needin' a comfortable place to lay anchor! Be sure ye get a sturdy cage, at least double the size of the bird's wingspan, with enough space to spread their wings, yarrrr!\n\n2. Perches 'n toys: Aye, parrots need perches of different sizes, shapes, 'n textures to keep their feet healthy. Also, a few toys be helpin' to keep them entertained 'n their minds stimulated, arrrh!\n\n3. Proper grub: Feed yer feathered friend a balanced diet of high-quality pellets, fruits, 'n veggies to keep 'em strong 'n healthy. Give 'em fresh water every day, or ye\u00e2\u20ac\u2122ll have a scurvy bird on yer hands!\n\n4. Cleanliness: Swab their cage deck! Clean their cage on a regular basis: fresh water 'n food daily, the floor every couple of days, 'n a thorough scrubbing ev'ry few weeks, so the bird be livin' in a tidy haven, arrhh!\n\n5. Socialize 'n train: Parrots be a sociable lot, arrr! Exercise 'n interact with 'em daily to create a bond 'n maintain their mental 'n physical health. Train 'em with positive reinforcement, treat 'em kindly, yarrr!\n\n6. Proper rest: Yer parrot be needin' \u00e2\u20ac\u2122bout 10-12 hours o' sleep each night. Cover their cage 'n let them slumber in a dim, quiet quarter for a proper night's rest, ye scallywag!\n\n7. Keep a weather eye open for illness: Birds be hidin' their ailments, arrr! Be watchful for signs of sickness, such as lethargy, loss of appetite, puffin' up, or change in droppings, and make haste to a vet if need be.\n\n8. Provide fresh air 'n avoid toxins: Parrots be sensitive to draft and pollutants. Keep yer quarters well ventilated, but no drafts, arrr! Be mindful of toxins like Teflon fumes, candles, or air fresheners.\n\nSo there ye have it, me hearty! With proper care 'n commitment, yer parrot will be squawkin' \"Yo-ho-ho\" for many years to come! Good luck, sailor, and may the wind be at yer back!"
        }
      }
    ],
    "usage": {
      "completion_tokens": 557,
      "prompt_tokens": 33,
      "total_tokens": 590
    }
  }
}

Example

Creates a completion based on Azure Search data and system-assigned managed identity.

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-06-01

{
 "messages": [
  {
   "role": "user",
   "content": "can you tell me how to care for a dog?"
  }
 ],
 "data_sources": [
  {
   "type": "azure_search",
   "parameters": {
    "endpoint": "https://your-search-endpoint.search.windows.net/",
    "index_name": "{index name}",
    "authentication": {
     "type": "system_assigned_managed_identity"
    }
   }
  }
 ]
}

Responses: Status Code: 200

{
  "body": {
    "id": "chatcmpl-7R1nGnsXO8n4oi9UPz2f3UHdgAYMn",
    "created": 1686676106,
    "choices": [
      {
        "index": 0,
        "finish_reason": "stop",
        "message": {
          "role": "assistant",
          "content": "Content of the completion [doc1].",
          "context": {
            "citations": [
              {
                "content": "Citation content.",
                "title": "Citation Title",
                "filepath": "contoso.txt",
                "url": "https://contoso.blob.windows.net/container/contoso.txt",
                "chunk_id": "0"
              }
            ],
            "intent": "dog care"
          }
        }
      }
    ],
    "usage": {
      "completion_tokens": 557,
      "prompt_tokens": 33,
      "total_tokens": 590
    }
  }
}

Example

Creates a completion based on Azure Search vector data, previous assistant message and user-assigned managed identity.

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-06-01

{
 "messages": [
  {
   "role": "user",
   "content": "can you tell me how to care for a cat?"
  },
  {
   "role": "assistant",
   "content": "Content of the completion [doc1].",
   "context": {
    "intent": "cat care"
   }
  },
  {
   "role": "user",
   "content": "how about dog?"
  }
 ],
 "data_sources": [
  {
   "type": "azure_search",
   "parameters": {
    "endpoint": "https://your-search-endpoint.search.windows.net/",
    "authentication": {
     "type": "user_assigned_managed_identity",
     "managed_identity_resource_id": "/subscriptions/{subscription-id}/resourceGroups/{resource-group}/providers/Microsoft.ManagedIdentity/userAssignedIdentities/{resource-name}"
    },
    "index_name": "{index name}",
    "query_type": "vector",
    "embedding_dependency": {
     "type": "deployment_name",
     "deployment_name": "{embedding deployment name}"
    },
    "in_scope": true,
    "top_n_documents": 5,
    "strictness": 3,
    "role_information": "you're an AI assistant that helps people find information.",
    "fields_mapping": {
     "content_fields_separator": "\\n",
     "content_fields": [
      "content"
     ],
     "filepath_field": "filepath",
     "title_field": "title",
     "url_field": "url",
     "vector_fields": [
      "contentvector"
     ]
    }
   }
  }
 ]
}

Responses: Status Code: 200

{
  "body": {
    "id": "chatcmpl-7R1nGnsXO8n4oi9UPz2f3UHdgAYMn",
    "created": 1686676106,
    "choices": [
      {
        "index": 0,
        "finish_reason": "stop",
        "message": {
          "role": "assistant",
          "content": "Content of the completion [doc1].",
          "context": {
            "citations": [
              {
                "content": "Citation content 2.",
                "title": "Citation Title 2",
                "filepath": "contoso2.txt",
                "url": "https://contoso.blob.windows.net/container/contoso2.txt",
                "chunk_id": "0"
              }
            ],
            "intent": "dog care"
          }
        }
      }
    ],
    "usage": {
      "completion_tokens": 557,
      "prompt_tokens": 33,
      "total_tokens": 590
    }
  }
}

Example

Creates a completion for the provided Azure Cosmos DB.

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-06-01

{
 "messages": [
  {
   "role": "user",
   "content": "can you tell me how to care for a dog?"
  }
 ],
 "data_sources": [
  {
   "type": "azure_cosmos_db",
   "parameters": {
    "authentication": {
     "type": "connection_string",
     "connection_string": "mongodb+srv://rawantest:{password}$@{cluster-name}.mongocluster.cosmos.azure.com/?tls=true&authMechanism=SCRAM-SHA-256&retrywrites=false&maxIdleTimeMS=120000"
    },
    "database_name": "vectordb",
    "container_name": "azuredocs",
    "index_name": "azuredocindex",
    "embedding_dependency": {
     "type": "deployment_name",
     "deployment_name": "{embedding deployment name}"
    },
    "fields_mapping": {
     "content_fields": [
      "content"
     ],
     "vector_fields": [
      "contentvector"
     ]
    }
   }
  }
 ]
}

Responses: Status Code: 200

{
  "body": {
    "id": "chatcmpl-7R1nGnsXO8n4oi9UPz2f3UHdgAYMn",
    "created": 1686676106,
    "choices": [
      {
        "index": 0,
        "finish_reason": "stop",
        "message": {
          "role": "assistant",
          "content": "Content of the completion [doc1].",
          "context": {
            "citations": [
              {
                "content": "Citation content.",
                "title": "Citation Title",
                "filepath": "contoso.txt",
                "url": "https://contoso.blob.windows.net/container/contoso.txt",
                "chunk_id": "0"
              }
            ],
            "intent": "dog care"
          }
        }
      }
    ],
    "usage": {
      "completion_tokens": 557,
      "prompt_tokens": 33,
      "total_tokens": 590
    }
  }
}

Transcriptions

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/transcriptions?api-version=2024-06-01

Transcribes audio into the input language.

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string	Deployment id of the whisper model.
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: multipart/form-data

Name	Type	Description	Required	Default
file	string	The audio file object to transcribe.	Yes
prompt	string	An optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.	No
response_format	audioResponseFormat	Defines the format of the output.	No
temperature	number	The sampling temperature, between 0 and 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. If set to 0, the model will use log probability to automatically increase the temperature until certain thresholds are hit.	No	0
language	string	The language of the input audio. Supplying the input language in ISO-639-1 format will improve accuracy and latency.	No

Responses

Status Code: 200

Description: OK

Content-Type	Type	Description
application/json	audioResponse or audioVerboseResponse
text/plain	string	Transcribed text in the output format (when response_format was one of text, vtt or srt).

Examples

Example

Gets transcribed text and associated metadata from provided spoken audio data.

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/transcriptions?api-version=2024-06-01

Responses: Status Code: 200

{
  "body": {
    "text": "A structured object when requesting json or verbose_json"
  }
}

Example

Gets transcribed text and associated metadata from provided spoken audio data.

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/transcriptions?api-version=2024-06-01

"---multipart-boundary\nContent-Disposition: form-data; name=\"file\"; filename=\"file.wav\"\nContent-Type: application/octet-stream\n\nRIFF..audio.data.omitted\n---multipart-boundary--"

Responses: Status Code: 200

{
  "type": "string",
  "example": "plain text when requesting text, srt, or vtt"
}

Translations

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/translations?api-version=2024-06-01

Transcribes and translates input audio into English text.

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string	Deployment id of the whisper model which was deployed.
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: multipart/form-data

Name	Type	Description	Required	Default
file	string	The audio file to translate.	Yes
prompt	string	An optional text to guide the model's style or continue a previous audio segment. The prompt should be in English.	No
response_format	audioResponseFormat	Defines the format of the output.	No
temperature	number	The sampling temperature, between 0 and 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. If set to 0, the model will use log probability to automatically increase the temperature until certain thresholds are hit.	No	0

Responses

Status Code: 200

Description: OK

Content-Type	Type	Description
application/json	audioResponse or audioVerboseResponse
text/plain	string	Transcribed text in the output format (when response_format was one of text, vtt or srt).

Examples

Example

Gets English language transcribed text and associated metadata from provided spoken audio data.

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/translations?api-version=2024-06-01

"---multipart-boundary\nContent-Disposition: form-data; name=\"file\"; filename=\"file.wav\"\nContent-Type: application/octet-stream\n\nRIFF..audio.data.omitted\n---multipart-boundary--"

Responses: Status Code: 200

{
  "body": {
    "text": "A structured object when requesting json or verbose_json"
  }
}

Example

Gets English language transcribed text and associated metadata from provided spoken audio data.

POST https://{endpoint}/openai/deployments/{deployment-id}/audio/translations?api-version=2024-06-01

"---multipart-boundary\nContent-Disposition: form-data; name=\"file\"; filename=\"file.wav\"\nContent-Type: application/octet-stream\n\nRIFF..audio.data.omitted\n---multipart-boundary--"

Responses: Status Code: 200

{
  "type": "string",
  "example": "plain text when requesting text, srt, or vtt"
}

Image generation

POST https://{endpoint}/openai/deployments/{deployment-id}/images/generations?api-version=2024-06-01

Generates a batch of images from a text caption on a given DALLE model deployment

URI Parameters

Name	In	Required	Type	Description
endpoint	path	Yes	string url	Supported Azure OpenAI endpoints (protocol and hostname, for example: `https://aoairesource.openai.azure.com`. Replace "aoairesource" with your Azure OpenAI resource name). https://{your-resource-name}.openai.azure.com
deployment-id	path	Yes	string	Deployment id of the `dall-e` model which was deployed.
api-version	query	Yes	string	API version

Request Header

Name	Required	Type	Description
api-key	True	string	Provide Azure OpenAI API key here

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
prompt	string	A text description of the desired image(s). The maximum length is 4000 characters.	Yes
n	integer	The number of images to generate.	No	1
size	imageSize	The size of the generated images.	No	1024x1024
response_format	imagesResponseFormat	The format in which the generated images are returned.	No	url
user	string	A unique identifier representing your end-user, which can help to monitor and detect abuse.	No
quality	imageQuality	The quality of the image that will be generated.	No	standard
style	imageStyle	The style of the generated images.	No	vivid

Responses

Status Code: 200

Description: Ok

Content-Type	Type	Description
application/json	generateImagesResponse

Status Code: default

Description: An error occurred.

Content-Type	Type	Description
application/json	dalleErrorResponse

Examples

Example

Creates images given a prompt.

POST https://{endpoint}/openai/deployments/{deployment-id}/images/generations?api-version=2024-06-01

{
 "prompt": "In the style of WordArt, Microsoft Clippy wearing a cowboy hat.",
 "n": 1,
 "style": "natural",
 "quality": "standard"
}

Responses: Status Code: 200

{
  "body": {
    "created": 1698342300,
    "data": [
      {
        "revised_prompt": "A vivid, natural representation of Microsoft Clippy wearing a cowboy hat.",
        "prompt_filter_results": {
          "sexual": {
            "severity": "safe",
            "filtered": false
          },
          "violence": {
            "severity": "safe",
            "filtered": false
          },
          "hate": {
            "severity": "safe",
            "filtered": false
          },
          "self_harm": {
            "severity": "safe",
            "filtered": false
          },
          "profanity": {
            "detected": false,
            "filtered": false
          }
        },
        "url": "https://dalletipusw2.blob.core.windows.net/private/images/e5451cc6-b1ad-4747-bd46-b89a3a3b8bc3/generated_00.png?se=2023-10-27T17%3A45%3A09Z&...",
        "content_filter_results": {
          "sexual": {
            "severity": "safe",
            "filtered": false
          },
          "violence": {
            "severity": "safe",
            "filtered": false
          },
          "hate": {
            "severity": "safe",
            "filtered": false
          },
          "self_harm": {
            "severity": "safe",
            "filtered": false
          }
        }
      }
    ]
  }
}

Components

errorResponse

Name	Type	Description	Required	Default
error	error		No

errorBase

Name	Type	Description	Required	Default
code	string		No
message	string		No

Error

Name	Type	Description	Required
param	string		No
type	string		No
inner_error	innerError	Inner error with additional details.	No

innerError

Inner error with additional details.

Name	Type	Description	Required	Default
code	innerErrorCode	Error codes for the inner error object.	No
content_filter_results	contentFilterPromptResults	Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about jailbreak content and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.	No

innerErrorCode

Error codes for the inner error object.

Description: Error codes for the inner error object.

Type: string

Default:

Enum Name: InnerErrorCode

Enum Values:

Value	Description
ResponsibleAIPolicyViolation	The prompt violated one of more content filter rules.

dalleErrorResponse

Name	Type	Description	Required	Default
error	dalleError		No

dalleError

Name	Type	Description	Required
param	string		No
type	string		No
inner_error	dalleInnerError	Inner error with additional details.	No

dalleInnerError

Inner error with additional details.

Name	Type	Description	Required
code	innerErrorCode	Error codes for the inner error object.	No
content_filter_results	dalleFilterResults	Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about jailbreak content and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.	No
revised_prompt	string	The prompt that was used to generate the image, if there was any revision to the prompt.	No

contentFilterResultBase

Name	Type	Description	Required	Default
filtered	boolean		Yes

contentFilterSeverityResult

Name	Type	Description	Required	Default
filtered	boolean		Yes
severity	string		No

contentFilterDetectedResult

Name	Type	Description	Required	Default
filtered	boolean		Yes
detected	boolean		No

contentFilterDetectedWithCitationResult

Name	Type	Description	Required	Default
citation	object		No

Properties for citation

URL

Name	Type	Description	Default
URL	string

License

Name	Type	Description	Default
license	string

contentFilterResultsBase

Information about the content filtering results.

Name	Type	Required
sexual	contentFilterSeverityResult	No
violence	contentFilterSeverityResult	No
hate	contentFilterSeverityResult	No
self_harm	contentFilterSeverityResult	No
profanity	contentFilterDetectedResult	No
error	errorBase	No

contentFilterPromptResults

Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about jailbreak content and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.

Name	Type	Required
sexual	contentFilterSeverityResult	No
violence	contentFilterSeverityResult	No
hate	contentFilterSeverityResult	No
self_harm	contentFilterSeverityResult	No
profanity	contentFilterDetectedResult	No
error	errorBase	No
jailbreak	contentFilterDetectedResult	No

contentFilterChoiceResults

Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about third-party text and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.

Name	Type	Required
sexual	contentFilterSeverityResult	No
violence	contentFilterSeverityResult	No
hate	contentFilterSeverityResult	No
self_harm	contentFilterSeverityResult	No
profanity	contentFilterDetectedResult	No
error	errorBase	No
protected_material_text	contentFilterDetectedResult	No
protected_material_code	contentFilterDetectedWithCitationResult	No

promptFilterResult

Content filtering results for a single prompt in the request.

Name	Type	Description	Required	Default
prompt_index	integer		No
content_filter_results	contentFilterPromptResults	Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about jailbreak content and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.	No

promptFilterResults

Content filtering results for zero or more prompts in the request. In a streaming request, results for different prompts may arrive at different times or in different orders.

dalleContentFilterResults

Information about the content filtering results.

Name	Type	Required
sexual	contentFilterSeverityResult	No
violence	contentFilterSeverityResult	No
hate	contentFilterSeverityResult	No
self_harm	contentFilterSeverityResult	No

dalleFilterResults

Name	Type	Required
sexual	contentFilterSeverityResult	No
violence	contentFilterSeverityResult	No
hate	contentFilterSeverityResult	No
self_harm	contentFilterSeverityResult	No
profanity	contentFilterDetectedResult	No
jailbreak	contentFilterDetectedResult	No

chatCompletionsRequestCommon

Name	Type	Description	Required	Default
temperature	number	What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. We generally recommend altering this or `top_p` but not both.	No	1
top_p	number	An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or `temperature` but not both.	No	1
stream	boolean	If set, partial message deltas will be sent, like in ChatGPT. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a `data: [DONE]` message.	No	False
stop	string or array	Up to 4 sequences where the API will stop generating further tokens.	No
max_tokens	integer	The maximum number of tokens allowed for the generated answer. By default, the number of tokens the model can return will be (4096 - prompt tokens).	No	4096
presence_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.	No	0
frequency_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.	No	0
logit_bias	object	Modify the likelihood of specified tokens appearing in the completion. Accepts a json object that maps tokens (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token.	No
user	string	A unique identifier representing your end-user, which can help Azure OpenAI to monitor and detect abuse.	No

createChatCompletionRequest

Name	Type	Description	Required	Default
temperature	number	What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. We generally recommend altering this or `top_p` but not both.	No	1
top_p	number	An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or `temperature` but not both.	No	1
stream	boolean	If set, partial message deltas will be sent, like in ChatGPT. Tokens will be sent as data-only server-sent events as they become available, with the stream terminated by a `data: [DONE]` message.	No	False
stop	string or array	Up to 4 sequences where the API will stop generating further tokens.	No
max_tokens	integer	The maximum number of tokens allowed for the generated answer. By default, the number of tokens the model can return will be (4096 - prompt tokens).	No	4096
presence_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.	No	0
frequency_penalty	number	Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.	No	0
logit_bias	object	Modify the likelihood of specified tokens appearing in the completion. Accepts a json object that maps tokens (specified by their token ID in the tokenizer) to an associated bias value from -100 to 100. Mathematically, the bias is added to the logits generated by the model prior to sampling. The exact effect will vary per model, but values between -1 and 1 should decrease or increase likelihood of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token.	No
user	string	A unique identifier representing your end-user, which can help Azure OpenAI to monitor and detect abuse.	No
messages	array	A list of messages comprising the conversation so far. Example Python code.	No
data_sources	array	The configuration entries for Azure OpenAI chat extensions that use them. This additional specification is only compatible with Azure OpenAI.	No
n	integer	How many chat completion choices to generate for each input message.	No	1
seed	integer	If specified, our system will make a best effort to sample deterministically, such that repeated requests with the same `seed` and parameters should return the same result. Determinism isn'tguaranteed, and you should refer to the `system_fingerprint` response parameter to monitor changes in the backend.	No	0
logprobs	boolean	Whether to return log probabilities of the output tokens or not. If true, returns the log probabilities of each output token returned in the `content` of `message`. This option is currently not available on the `gpt-4-vision-preview` model.	No	False
top_logprobs	integer	An integer between 0 and 5 specifying the number of most likely tokens to return at each token position, each with an associated log probability. `logprobs` must be set to `true` if this parameter is used.	No
response_format	object	An object specifying the format that the model must output. Used to enable JSON mode.	No
tools	array	A list of tools the model may call. Currently, only functions are supported as a tool. Use this to provide a list of functions the model may generate JSON inputs for.	No
tool_choice	chatCompletionToolChoiceOption	Controls which (if any) function is called by the model. `none` means the model will not call a function and instead generates a message. `auto` means the model can pick between generating a message or calling a function. Specifying a particular function via `{"type": "function", "function": {"name": "my_function"}}` forces the model to call that function.	No
functions	array	Deprecated in favor of `tools`. A list of functions the model may generate JSON inputs for.	No
function_call	string or object	Deprecated in favor of `tool_choice`. Controls how the model responds to function calls. "none" means the model doesn't call a function, and responds to the end-user. "auto" means the model can pick between an end-user or calling a function. Specifying a particular function via `{"name":\ "my_function"}` forces the model to call that function. "none" is the default when no functions are present. "auto" is the default if functions are present.	No

Properties for response_format

Type

Name	Type	Description	Default
type	chatCompletionResponseFormat	Setting to `json_object` enables JSON mode. This guarantees that the message the model generates is valid JSON.	text

chatCompletionResponseFormat

Setting to json_object enables JSON mode. This guarantees that the message the model generates is valid JSON.

Description: Setting to json_object enables JSON mode. This guarantees that the message the model generates is valid JSON.

Type: string

Default: text

Enum Name: ChatCompletionResponseFormat

Enum Values:

Value	Description
text	Response format is a plain text string.
json_object	Response format is a JSON object.

chatCompletionFunction

Name	Type	Description	Required
name	string	The name of the function to be called. Must be a-z, A-Z, 0-9, or contain underscores and dashes, with a maximum length of 64.	Yes
description	string	The description of what the function does.	No
parameters	chatCompletionFunctionParameters	The parameters the functions accepts, described as a JSON Schema object. See the JSON Schema reference for documentation about the format.	No

chatCompletionFunctionParameters

The parameters the functions accepts, described as a JSON Schema object. See the JSON Schema reference for documentation about the format.

chatCompletionRequestMessage

Name	Type	Description	Required	Default
role	chatCompletionRequestMessageRole	The role of the messages author.	Yes

chatCompletionRequestMessageRole

The role of the messages author.

Description: The role of the messages author.

Type: string

Default:

Enum Name: ChatCompletionRequestMessageRole

Enum Values:

Value	Description
system	The message author role is system.
user	The message author role is user.
assistant	The message author role is assistant.
tool	The message author role is tool.
function	Deprecated. The message author role is function.

chatCompletionRequestMessageSystem

Name	Type	Description	Required	Default
role	chatCompletionRequestMessageRole	The role of the messages author.	Yes
content	string	The contents of the message.	No

chatCompletionRequestMessageUser

Name	Type	Description	Required	Default
role	chatCompletionRequestMessageRole	The role of the messages author.	Yes
content	string or array		No

chatCompletionRequestMessageContentPart

Name	Type	Description	Required	Default
type	chatCompletionRequestMessageContentPartType	The type of the content part.	Yes

chatCompletionRequestMessageContentPartType

The type of the content part.

Description: The type of the content part.

Type: string

Default:

Enum Name: ChatCompletionRequestMessageContentPartType

Enum Values:

Value	Description
text	The content part type is text.
image_url	The content part type is image_url.

chatCompletionRequestMessageContentPartText

Name	Type	Description	Required	Default
type	chatCompletionRequestMessageContentPartType	The type of the content part.	Yes
text	string	The text content.	No

chatCompletionRequestMessageContentPartImage

Name	Type	Description	Required	Default
type	chatCompletionRequestMessageContentPartType	The type of the content part.	Yes
url	string	Either a URL of the image or the base64 encoded image data.	No
detail	imageDetailLevel	Specifies the detail level of the image.	No	auto

imageDetailLevel

Specifies the detail level of the image.

Description: Specifies the detail level of the image.

Type: string

Default: auto

Enum Name: ImageDetailLevel

Enum Values:

Value	Description
auto	The image detail level is auto.
low	The image detail level is low.
high	The image detail level is high.

chatCompletionRequestMessageAssistant

Name	Type	Description	Required
role	chatCompletionRequestMessageRole	The role of the messages author.	Yes
content	string	The contents of the message.	No
tool_calls	array	The tool calls generated by the model, such as function calls.	No
context	azureChatExtensionsMessageContext	A representation of the additional context information available when Azure OpenAI chat extensions are involved in the generation of a corresponding chat completions response. This context information is only populated when using an Azure OpenAI request configured to use a matching extension.	No

azureChatExtensionConfiguration

A representation of configuration data for a single Azure OpenAI chat extension. This will be used by a chat completions request that should use Azure OpenAI chat extensions to augment the response behavior. The use of this configuration is compatible only with Azure OpenAI.

Name	Type	Description	Required	Default
type	azureChatExtensionType	A representation of configuration data for a single Azure OpenAI chat extension. This will be used by a chat completions request that should use Azure OpenAI chat extensions to augment the response behavior. The use of this configuration is compatible only with Azure OpenAI.	Yes

azureChatExtensionType

Description: A representation of configuration data for a single Azure OpenAI chat extension. This will be used by a chat
Completions request that should use Azure OpenAI chat extensions to augment the response behavior.
The use of this configuration is compatible only with Azure OpenAI.

Type: string

Default:

Enum Name: AzureChatExtensionType

Enum Values:

Value	Description
azure_search	Represents the use of Azure Search as an Azure OpenAI chat extension.
azure_cosmos_db	Represents the use of Azure Cosmos DB as an Azure OpenAI chat extension.

azureSearchChatExtensionConfiguration

A specific representation of configurable options for Azure Search when using it as an Azure OpenAI chat extension.

Name	Type	Description	Required	Default
type	azureChatExtensionType	A representation of configuration data for a single Azure OpenAI chat extension. This will be used by a chat completions request that should use Azure OpenAI chat extensions to augment the response behavior. The use of this configuration is compatible only with Azure OpenAI.	Yes
parameters	azureSearchChatExtensionParameters	Parameters for Azure Search when used as an Azure OpenAI chat extension.	No

azureSearchChatExtensionParameters

Parameters for Azure Search when used as an Azure OpenAI chat extension.

Name	Type	Description	Required
authentication	onYourDataApiKeyAuthenticationOptions or onYourDataSystemAssignedManagedIdentityAuthenticationOptions or onYourDataUserAssignedManagedIdentityAuthenticationOptions		Yes
top_n_documents	integer	The configured top number of documents to feature for the configured query.	No
in_scope	boolean	Whether queries should be restricted to use of indexed data.	No
strictness	integer	The configured strictness of the search relevance filtering. The higher of strictness, the higher of the precision but lower recall of the answer.	No
role_information	string	Give the model instructions about how it should behave and any context it should reference when generating a response. You can describe the assistant's personality and tell it how to format responses. There's a 100 token limit for it, and it counts against the overall token limit.	No
endpoint	string	The absolute endpoint path for the Azure Search resource to use.	Yes
index_name	string	The name of the index to use as available in the referenced Azure Search resource.	Yes
fields_mapping	azureSearchIndexFieldMappingOptions	Optional settings to control how fields are processed when using a configured Azure Search resource.	No
query_type	azureSearchQueryType	The type of Azure Search retrieval query that should be executed when using it as an Azure OpenAI chat extension.	No
semantic_configuration	string	The additional semantic configuration for the query.	No
filter	string	Search filter.	No
embedding_dependency	onYourDataEndpointVectorizationSource or onYourDataDeploymentNameVectorizationSource		No

azureSearchIndexFieldMappingOptions

Optional settings to control how fields are processed when using a configured Azure Search resource.

Name	Type	Description	Required
title_field	string	The name of the index field to use as a title.	No
url_field	string	The name of the index field to use as a URL.	No
filepath_field	string	The name of the index field to use as a filepath.	No
content_fields	array	The names of index fields that should be treated as content.	No
content_fields_separator	string	The separator pattern that content fields should use.	No
vector_fields	array	The names of fields that represent vector data.	No

azureSearchQueryType

The type of Azure Search retrieval query that should be executed when using it as an Azure OpenAI chat extension.

Description: The type of Azure Search retrieval query that should be executed when using it as an Azure OpenAI chat extension.

Type: string

Default:

Enum Name: azureSearchQueryType

Enum Values:

Value	Description
simple	Represents the default, simple query parser.
semantic	Represents the semantic query parser for advanced semantic modeling.
vector	Represents vector search over computed data.
vector_simple_hybrid	Represents a combination of the simple query strategy with vector data.
vector_semantic_hybrid	Represents a combination of semantic search and vector data querying.

azureCosmosDBChatExtensionConfiguration

A specific representation of configurable options for Azure Cosmos DB when using it as an Azure OpenAI chat extension.

Name	Type	Description	Required	Default
type	azureChatExtensionType	A representation of configuration data for a single Azure OpenAI chat extension. This will be used by a chat completions request that should use Azure OpenAI chat extensions to augment the response behavior. The use of this configuration is compatible only with Azure OpenAI.	Yes
parameters	azureCosmosDBChatExtensionParameters	Parameters to use when configuring Azure OpenAI On Your Data chat extensions when using Azure Cosmos DB for MongoDB vCore.	No

azureCosmosDBChatExtensionParameters

Parameters to use when configuring Azure OpenAI On Your Data chat extensions when using Azure Cosmos DB for MongoDB vCore.

Name	Type	Description	Required
authentication	onYourDataConnectionStringAuthenticationOptions	The authentication options for Azure OpenAI On Your Data when using a connection string.	Yes
top_n_documents	integer	The configured top number of documents to feature for the configured query.	No
in_scope	boolean	Whether queries should be restricted to use of indexed data.	No
strictness	integer	The configured strictness of the search relevance filtering. The higher of strictness, the higher of the precision but lower recall of the answer.	No
role_information	string	Give the model instructions about how it should behave and any context it should reference when generating a response. You can describe the assistant's personality and tell it how to format responses. There's a 100 token limit for it, and it counts against the overall token limit.	No
database_name	string	The MongoDB vCore database name to use with Azure Cosmos DB.	Yes
container_name	string	The name of the Azure Cosmos DB resource container.	Yes
index_name	string	The MongoDB vCore index name to use with Azure Cosmos DB.	Yes
fields_mapping	azureCosmosDBFieldMappingOptions	Optional settings to control how fields are processed when using a configured Azure Cosmos DB resource.	Yes
embedding_dependency	onYourDataEndpointVectorizationSource or onYourDataDeploymentNameVectorizationSource		Yes

azureCosmosDBFieldMappingOptions

Optional settings to control how fields are processed when using a configured Azure Cosmos DB resource.

Name	Type	Description	Required
title_field	string	The name of the index field to use as a title.	No
url_field	string	The name of the index field to use as a URL.	No
filepath_field	string	The name of the index field to use as a filepath.	No
content_fields	array	The names of index fields that should be treated as content.	Yes
content_fields_separator	string	The separator pattern that content fields should use.	No
vector_fields	array	The names of fields that represent vector data.	Yes

onYourDataAuthenticationOptions

The authentication options for Azure OpenAI On Your Data.

Name	Type	Description	Required	Default
type	onYourDataAuthenticationType	The authentication types supported with Azure OpenAI On Your Data.	Yes

onYourDataAuthenticationType

The authentication types supported with Azure OpenAI On Your Data.

Description: The authentication types supported with Azure OpenAI On Your Data.

Type: string

Default:

Enum Name: OnYourDataAuthenticationType

Enum Values:

Value	Description
api_key	Authentication via API key.
connection_string	Authentication via connection string.
system_assigned_managed_identity	Authentication via system-assigned managed identity.
user_assigned_managed_identity	Authentication via user-assigned managed identity.

onYourDataApiKeyAuthenticationOptions

The authentication options for Azure OpenAI On Your Data when using an API key.

Name	Type	Description	Required	Default
type	onYourDataAuthenticationType	The authentication types supported with Azure OpenAI On Your Data.	Yes
key	string	The API key to use for authentication.	No

onYourDataConnectionStringAuthenticationOptions

The authentication options for Azure OpenAI On Your Data when using a connection string.

Name	Type	Description	Required	Default
type	onYourDataAuthenticationType	The authentication types supported with Azure OpenAI On Your Data.	Yes
connection_string	string	The connection string to use for authentication.	No

onYourDataSystemAssignedManagedIdentityAuthenticationOptions

The authentication options for Azure OpenAI On Your Data when using a system-assigned managed identity.

Name	Type	Description	Required	Default
type	onYourDataAuthenticationType	The authentication types supported with Azure OpenAI On Your Data.	Yes

onYourDataUserAssignedManagedIdentityAuthenticationOptions

The authentication options for Azure OpenAI On Your Data when using a user-assigned managed identity.

Name	Type	Description	Required	Default
type	onYourDataAuthenticationType	The authentication types supported with Azure OpenAI On Your Data.	Yes
managed_identity_resource_id	string	The resource ID of the user-assigned managed identity to use for authentication.	No

onYourDataVectorizationSource

An abstract representation of a vectorization source for Azure OpenAI On Your Data with vector search.

Name	Type	Description	Required	Default
type	onYourDataVectorizationSourceType	Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with vector search.	Yes

onYourDataVectorizationSourceType

Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with vector search.

Description: Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with
Vector search.

Type: string

Default:

Enum Name: OnYourDataVectorizationSourceType

Enum Values:

Value	Description
endpoint	Represents vectorization performed by public service calls to an Azure OpenAI embedding model.
deployment_name	Represents an Ada model deployment name to use. This model deployment must be in the same Azure OpenAI resource, but The on your data feature will use this model deployment via an internal call rather than a public one, which enables vector search even in private networks.

onYourDataDeploymentNameVectorizationSource

The details of a vectorization source, used by Azure OpenAI On Your Data when applying vector search, that is based on an internal embeddings model deployment name in the same Azure OpenAI resource.

Name	Type	Description	Required	Default
type	onYourDataVectorizationSourceType	Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with vector search.	Yes
deployment_name	string	Specifies the name of the model deployment to use for vectorization. This model deployment must be in the same Azure OpenAI resource, but On Your Data will use this model deployment via an internal call rather than a public one, which enables vector search even in private networks.	No

onYourDataEndpointVectorizationSource

The details of a vectorization source, used by Azure OpenAI On Your Data when applying vector search, that is based on public embeddings endpoint for Azure OpenAI.

Name	Type	Description	Required
type	onYourDataVectorizationSourceType	Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with vector search.	Yes
authentication	onYourDataApiKeyAuthenticationOptions	The authentication options for Azure OpenAI On Your Data when using an API key.	No
endpoint	string	Specifies the endpoint to use for vectorization. This endpoint must be in the same Azure OpenAI resource, but On Your Data will use this endpoint via an internal call rather than a public one, which enables vector search even in private networks.	No

azureChatExtensionsMessageContext

A representation of the additional context information available when Azure OpenAI chat extensions are involved in the generation of a corresponding chat completions response. This context information is only populated when using an Azure OpenAI request configured to use a matching extension.

Name	Type	Description	Required	Default
citations	array	The data source retrieval result, used to generate the assistant message in the response.	No
intent	string	The detected intent from the chat history, used to pass to the next turn to carry over the context.	No

Citation

Citation information for a chat completions response message.

Name	Type	Description	Required
content	string	The content of the citation.	Yes
title	string	The title of the citation.	No
url	string	The URL of the citation.	No
filepath	string	The file path of the citation.	No
chunk_id	string	The chunk ID of the citation.	No

chatCompletionMessageToolCall

Name	Type	Description	Required
id	string	The ID of the tool call.	Yes
type	toolCallType	The type of the tool call, in this case `function`.	Yes
function	object	The function that the model called.	Yes

Properties for function

Name

Name	Type	Description	Default
name	string	The name of the function to call.

Arguments

Name	Type	Description	Default
arguments	string	The arguments to call the function with, as generated by the model in JSON format. Note that the model doesn't always generate valid JSON, and may fabricate parameters not defined by your function schema. Validate the arguments in your code before calling your function.

toolCallType

The type of the tool call, in this case function.

Description: The type of the tool call, in this case function.

Type: string

Default:

Enum Name: ToolCallType

Enum Values:

Value	Description
function	The tool call type is function.

chatCompletionRequestMessageTool

Name	Type	Description	Required
role	chatCompletionRequestMessageRole	The role of the messages author.	Yes
tool_call_id	string	Tool call that this message is responding to.	No
content	string	The contents of the message.	No

chatCompletionRequestMessageFunction

Name	Type	Description	Required
role	enum	The role of the messages author, in this case `function`. Possible values: function	Yes
name	string	The contents of the message.	No
content	string	The contents of the message.	No

createChatCompletionResponse

Name	Type	Description	Required
id	string	A unique identifier for the chat completion.	Yes
object	chatCompletionResponseObject	The object type.	Yes
created	integer	The Unix timestamp (in seconds) of when the chat completion was created.	Yes
model	string	The model used for the chat completion.	Yes
usage	completionUsage	Usage statistics for the completion request.	No
system_fingerprint	string	Can be used in conjunction with the `seed` request parameter to understand when backend changes have been made that might impact determinism.	No
prompt_filter_results	promptFilterResults	Content filtering results for zero or more prompts in the request. In a streaming request, results for different prompts may arrive at different times or in different orders.	No
choices	array		No

chatCompletionChoiceLogProbs

Log probability information for the choice.

Name	Type	Description	Required	Default
content	array	A list of message content tokens with log probability information.	Yes

chatCompletionTokenLogprob

Name	Type	Description	Required
token	string	The token.	Yes
logprob	number	The log probability of this token.	Yes
bytes	array	A list of integers representing the UTF-8 bytes representation of the token. Useful in instances where characters are represented by multiple tokens and their byte representations must be combined to generate the correct text representation. Can be `null` if there is no bytes representation for the token.	Yes
top_logprobs	array	List of the most likely tokens and their log probability, at this token position. In rare cases, there may be fewer than the number of requested `top_logprobs` returned.	Yes

chatCompletionResponseMessage

A chat completion message generated by the model.

Name	Type	Description	Required
role	chatCompletionResponseMessageRole	The role of the author of the response message.	No
content	string	The contents of the message.	No
tool_calls	array	The tool calls generated by the model, such as function calls.	No
function_call	chatCompletionFunctionCall	Deprecated and replaced by `tool_calls`. The name and arguments of a function that should be called, as generated by the model.	No
context	azureChatExtensionsMessageContext	A representation of the additional context information available when Azure OpenAI chat extensions are involved in the generation of a corresponding chat completions response. This context information is only populated when using an Azure OpenAI request configured to use a matching extension.	No

chatCompletionResponseMessageRole

The role of the author of the response message.

Description: The role of the author of the response message.

Type: string

Default:

Enum Values:

assistant

chatCompletionToolChoiceOption

Controls which (if any) function is called by the model. none means the model will not call a function and instead generates a message. auto means the model can pick between generating a message or calling a function. Specifying a particular function via {"type": "function", "function": {"name": "my_function"}} forces the model to call that function.

This component can be one of the following:

chatCompletionNamedToolChoice

Specifies a tool the model should use. Use to force the model to call a specific function.

Name	Type	Description	Required	Default
type	enum	The type of the tool. Currently, only `function` is supported. Possible values: function	No
function	object		No

Properties for function

Name

Name	Type	Description	Default
name	string	The name of the function to call.

chatCompletionFunctionCall

Deprecated and replaced by tool_calls. The name and arguments of a function that should be called, as generated by the model.

Name	Type	Description	Required	Default
name	string	The name of the function to call.	Yes
arguments	string	The arguments to call the function with, as generated by the model in JSON format. Note that the model doesn't always generate valid JSON, and may fabricate parameters not defined by your function schema. Validate the arguments in your code before calling your function.	Yes

chatCompletionsResponseCommon

Name	Type	Description	Required
id	string	A unique identifier for the chat completion.	Yes
object	chatCompletionResponseObject	The object type.	Yes
created	integer	The Unix timestamp (in seconds) of when the chat completion was created.	Yes
model	string	The model used for the chat completion.	Yes
usage	completionUsage	Usage statistics for the completion request.	No
system_fingerprint	string	Can be used in conjunction with the `seed` request parameter to understand when backend changes have been made that might impact determinism.	No

chatCompletionResponseObject

The object type.

Description: The object type.

Type: string

Default:

Enum Name: ChatCompletionResponseObject

Enum Values:

Value	Description
chat.completion	The object type is chat completion.

completionUsage

Usage statistics for the completion request.

Name	Type	Description	Required
prompt_tokens	integer	Number of tokens in the prompt.	Yes
completion_tokens	integer	Number of tokens in the generated completion.	Yes
total_tokens	integer	Total number of tokens used in the request (prompt + completion).	Yes

chatCompletionTool

Name	Type	Description	Required	Default
type	chatCompletionToolType	The type of the tool. Currently, only `function` is supported.	Yes
function	object		Yes

Properties for function

Description

Name	Type	Description	Default
description	string	A description of what the function does, used by the model to choose when and how to call the function.

Name

Name	Type	Description	Default
name	string	The name of the function to be called. Must be a-z, A-Z, 0-9, or contain underscores and dashes, with a maximum length of 64.

Parameters

Name	Type	Description	Default
parameters	chatCompletionFunctionParameters	The parameters the functions accepts, described as a JSON Schema object. See the JSON Schema reference for documentation about the format.

chatCompletionToolType

The type of the tool. Currently, only function is supported.

Description: The type of the tool. Currently, only function is supported.

Type: string

Default:

Enum Name: ChatCompletionToolType

Enum Values:

Value	Description
function	The tool type is function.

chatCompletionChoiceCommon

Name	Type	Description	Required	Default
index	integer		No
finish_reason	string		No

createTranslationRequest

Translation request.

Name	Type	Description	Required	Default
file	string	The audio file to translate.	Yes
prompt	string	An optional text to guide the model's style or continue a previous audio segment. The prompt should be in English.	No
response_format	audioResponseFormat	Defines the format of the output.	No
temperature	number	The sampling temperature, between 0 and 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. If set to 0, the model will use log probability to automatically increase the temperature until certain thresholds are hit.	No	0

audioResponse

Translation or transcription response when response_format was json

Name	Type	Description	Required	Default
text	string	Translated or transcribed text.	Yes

audioVerboseResponse

Translation or transcription response when response_format was verbose_json

Name	Type	Description	Required
text	string	Translated or transcribed text.	Yes
task	string	Type of audio task.	No
language	string	Language.	No
duration	number	Duration.	No
segments	array		No

audioResponseFormat

Defines the format of the output.

Description: Defines the format of the output.

Type: string

Default:

Enum Values:

json
text
srt
verbose_json
vtt

createTranscriptionRequest

Transcription request.

Name	Type	Description	Required	Default
file	string	The audio file object to transcribe.	Yes
prompt	string	An optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.	No
response_format	audioResponseFormat	Defines the format of the output.	No
temperature	number	The sampling temperature, between 0 and 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. If set to 0, the model will use log probability to automatically increase the temperature until certain thresholds are hit.	No	0
language	string	The language of the input audio. Supplying the input language in ISO-639-1 format will improve accuracy and latency.	No

audioSegment

Transcription or translation segment.

Name	Type	Description	Required
id	integer	Segment identifier.	No
seek	number	Offset of the segment.	No
start	number	The segment start offset.	No
end	number	Segment end offset.	No
text	string	Segment text.	No
tokens	array	Tokens of the text.	No
temperature	number	Temperature.	No
avg_logprob	number	Average log probability.	No
compression_ratio	number	Compression ratio.	No
no_speech_prob	number	Probability of 'no speech'.	No

imageQuality

The quality of the image that will be generated.

Description: The quality of the image that will be generated.

Type: string

Default: standard

Enum Name: Quality

Enum Values:

Value	Description
standard	Standard quality creates images with standard quality.
hd	HD quality creates images with finer details and greater consistency across the image.

imagesResponseFormat

The format in which the generated images are returned.

Description: The format in which the generated images are returned.

Type: string

Default: url

Enum Name: ImagesResponseFormat

Enum Values:

Value	Description
url	The URL that provides temporary access to download the generated images.
b64_json	The generated images are returned as base64 encoded string.

imageSize

The size of the generated images.

Description: The size of the generated images.

Type: string

Default: 1024x1024

Enum Name: Size

Enum Values:

Value	Description
1792x1024	The desired size of the generated image is 1792x1024 pixels.
1024x1792	The desired size of the generated image is 1024x1792 pixels.
1024x1024	The desired size of the generated image is 1024x1024 pixels.

imageStyle

The style of the generated images.

Description: The style of the generated images.

Type: string

Default: vivid

Enum Name: Style

Enum Values:

Value	Description
vivid	Vivid creates images that are hyper-realistic and dramatic.
natural	Natural creates images that are more natural and less hyper-realistic.

imageGenerationsRequest

Name	Type	Description	Required	Default
prompt	string	A text description of the desired image(s). The maximum length is 4000 characters.	Yes
n	integer	The number of images to generate.	No	1
size	imageSize	The size of the generated images.	No	1024x1024
response_format	imagesResponseFormat	The format in which the generated images are returned.	No	url
user	string	A unique identifier representing your end-user, which can help to monitor and detect abuse.	No
quality	imageQuality	The quality of the image that will be generated.	No	standard
style	imageStyle	The style of the generated images.	No	vivid

generateImagesResponse

Name	Type	Description	Required	Default
created	integer	The unix timestamp when the operation was created.	Yes
data	array	The result data of the operation, if successful	Yes

imageResult

The image url or encoded image if successful, and an error otherwise.

Name	Type	Description	Required
url	string	The image url.	No
b64_json	string	The base64 encoded image	No
content_filter_results	dalleContentFilterResults	Information about the content filtering results.	No
revised_prompt	string	The prompt that was used to generate the image, if there was any revision to the prompt.	No
prompt_filter_results	dalleFilterResults	Information about the content filtering category (hate, sexual, violence, self_harm), if it has been detected, as well as the severity level (very_low, low, medium, high-scale that determines the intensity and risk level of harmful content) and if it has been filtered or not. Information about jailbreak content and profanity, if it has been detected, and if it has been filtered or not. And information about customer blocklist, if it has been filtered and its id.	No

Completions extensions

Completions extensions aren't part of the latest GA version of the Azure OpenAI data plane inference spec.

Chatmessage

The Chat message object isn't part of the latest GA version of the Azure OpenAI data plane inference spec.

Text to speech

Is not currently part of the latest Azure OpenAI GA version of the Azure OpenAI data plane inference spec. Refer to the latest preview version for this capability.

Next steps

Learn about Models, and fine-tuning with the REST API. Learn more about the underlying models that power Azure OpenAI.

Share via

Azure OpenAI Service REST API reference

API specs

Authentication

REST API versioning

Data plane inference

Completions

URI Parameters

Request Header

Request Body

Responses

Properties for usage

completion_tokens

Examples

Example

Embeddings

URI Parameters

Request Header

Request Body

Responses

Properties for usage

prompt_tokens

total_tokens

Examples

Example

Chat completions

URI Parameters

Request Header

Request Body

Properties for response_format

Type

Responses

Examples

Example

Example

Example

Example

Transcriptions

URI Parameters

Request Header

Request Body

Responses

Examples

Example

Example

Translations

URI Parameters

Request Header

Request Body

Responses

Examples

Example

Example

Image generation

URI Parameters

Request Header

Request Body

Responses

Examples

Example

Components

errorResponse

errorBase

Error

innerError

innerErrorCode

dalleErrorResponse

dalleError

dalleInnerError

contentFilterResultBase

contentFilterSeverityResult

contentFilterDetectedResult

contentFilterDetectedWithCitationResult

Properties for citation

URL

License

contentFilterResultsBase

contentFilterPromptResults

contentFilterChoiceResults

promptFilterResult