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/chat/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, 2024-10-21.
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=2024-10-21
Creates a completion for the provided prompt, parameters, and chosen model.
URI Parameters
Required
Description
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
string
Deployment ID of the model which was deployed.
api-version
query
string
API version
Request Header
Required
Description
prompt
string or array
The prompt(s) to generate completions for, encoded as a string, array of strings, array of tokens, or array of token arrays.

Note that <|endoftext|> is the document separator that the model sees during training, so if a prompt isn't specified the model will generate as if from the beginning of a new document.

best_of
integer
Generates best_of completions server-side and returns the "best" (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.

boolean
Echo back the prompt in addition to the completion

False
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.

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 GPT 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.

As an example, you can pass {"50256": -100} to prevent the <|endoftext|> token from being generated.

logprobs
integer
Include the log probabilities on the logprobs most likely output tokens, as well the chosen tokens. For example, if logprobs is 5, the API will return a list of the five 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.

The maximum value for logprobs is 5.

max_tokens
integer
The maximum number of tokens that can be generated in the completion.

The token count of your prompt plus max_tokens can't exceed the model's context length.
integer
How many completions to generate for each prompt.

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.

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.

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't guaranteed, and you should refer to the system_fingerprint response parameter to monitor changes in the backend.

string or array
Up to four sequences where the API will stop generating further tokens. The returned text won't contain the stop sequence.

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.
False
suffix
string
The suffix that comes after a completion of inserted text.

This parameter is only supported for gpt-3.5-turbo-instruct.

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.

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.

string
A unique identifier representing your end-user, which can help to monitor and detect abuse.

Examples
Example
Creates a completion for the provided prompt, parameters, and chosen model.
POST https://{endpoint}/openai/deployments/{deployment-id}/completions?api-version=2024-10-21
 "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=2024-10-21
Get a vector representation of a given input that can be easily consumed by machine learning models and algorithms.
URI Parameters
Required
Description
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
string
api-version
query
string
API version
Request Header
Required
Description
input
string or array
Input text to embed, encoded as a string or array of tokens. To embed multiple inputs in a single request, pass an array of strings or array of token arrays. The input must not exceed the max input tokens for the model (8,192 tokens for text-embedding-ada-002), can't be an empty string, and any array must be 2,048 dimensions or less.
string
A unique identifier representing your end-user, which can help monitoring and detecting abuse.
input_type
string
input type of embedding search to use
encoding_format
string
The format to return the embeddings in. Can be either float or base64. Defaults to float.
dimensions
integer
The number of dimensions the resulting output embeddings should have. Only supported in text-embedding-3 and later models.
Example
Return the embeddings for a given prompt.
POST https://{endpoint}/openai/deployments/{deployment-id}/embeddings?api-version=2024-10-21
 "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,
          -0.00535031,
          0.008968075,
          0.014351576,
          -0.014086051,
          0.015055214,
          -0.022211088,
          -0.025198232,
          0.0065186154,
          -0.036350243,
          0.009180495,
          -0.009698266,
          0.009446018,
          -0.008463579,
          -0.0040426035,
          -0.03443847,
          -0.00091273896,
          -0.0019217303,
          0.002349888,
          -0.021560553,
          0.016515596,
          -0.015572986,
          0.0038666942,
          -8.432463e-05
    "usage": {
      "prompt_tokens": 4,
      "total_tokens": 4
Chat completions
POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-10-21
Creates a completion for the chat message
URI Parameters
Required
Description
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
string
Deployment ID of the model which was deployed.
api-version
query
string
API version
Request Header
Required
Description
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.

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.

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.
False
string or array
Up to four sequences where the API will stop generating further tokens.

max_tokens
integer
The maximum number of tokens that can be generated in the chat completion.

The total length of input tokens and generated tokens is limited by the model's context length.
max_completion_tokens
integer
An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.
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.

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.

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.

string
A unique identifier representing your end-user, which can help to monitor and detect abuse.

messages
array
A list of messages comprising the conversation so far.
data_sources
array
The configuration entries for Azure OpenAI chat extensions that use them.
  This additional specification is only compatible with Azure OpenAI.
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.
False
top_logprobs
integer
An integer between 0 and 20 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.
integer
How many chat completion choices to generate for each input message. Note that you'll be charged based on the number of generated tokens across all of the choices. Keep n as 1 to minimize costs.
parallel_tool_calls
ParallelToolCalls
Whether to enable parallel function calling during tool use.
response_format
ResponseFormatText or ResponseFormatJsonObject or ResponseFormatJsonSchema
An object specifying the format that the model must output. Compatible with GPT-4o, GPT-4o mini, GPT-4 Turbo and all GPT-3.5 Turbo models newer than gpt-3.5-turbo-1106.

Setting to { "type": "json_schema", "json_schema": {...} } enables Structured Outputs which guarantees the model will match your supplied JSON schema.

Setting to { "type": "json_object" } enables JSON mode, which guarantees the message the model generates is valid JSON.

Important: when using JSON mode, you must also instruct the model to produce JSON yourself via a system or user message. Without this, the model may generate an unending stream of whitespace until the generation reaches the token limit, resulting in a long-running and seemingly "stuck" request. Also note that the message content may be partially cut off if finish_reason="length", which indicates the generation exceeded max_tokens or the conversation exceeded the max context length.

integer
This feature is in Beta.
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't guaranteed, and you should refer to the system_fingerprint response parameter to monitor changes in the backend.

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. A max of 128 functions are supported.

tool_choice
chatCompletionToolChoiceOption
Controls which (if any) tool is called by the model. none means the model won't call any tool and instead generates a message. auto means the model can pick between generating a message or calling one or more tools. required means the model must call one or more tools. Specifying a particular tool via {"type": "function", "function": {"name": "my_function"}} forces the model to call that tool. none is the default when no tools are present. auto is the default if tools are present.
function_call
string or chatCompletionFunctionCallOption
Deprecated in favor of tool_choice.

Controls which (if any) function is called by the model.
none means the model won't 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 {"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.

functions
array
Deprecated in favor of tools.

A list of functions the model may generate JSON inputs for.

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




    

POST https://{endpoint}/openai/deployments/{deployment-id}/chat/completions?api-version=2024-10-21
 "messages": [
   "role": "system",
   "content": "you are 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=2024-10-21
 "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=2024-10-21
 "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 are 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=2024-10-21
 "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 - Create
POST https://{endpoint}/openai/deployments/{deployment-id}/audio/transcriptions?api-version=2024-10-21
Transcribes audio into the input language.
URI Parameters
Required
Description
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
string
Deployment ID of the speech to text model.

For information about supported models, see [/azure/ai-foundry/openai/concepts/models#audio-models].
api-version
query
string
API version
Request Header
Required
Description
prompt
string
An optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.
response_format
audioResponseFormat
Defines the format of the output.
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.
language
string
The language of the input audio. Supplying the input language in ISO-639-1 format will improve accuracy and latency.
Examples
Example
Gets transcribed text and associated metadata from provided spoken audio data.
POST https://{endpoint}/openai/deployments/{deployment-id}/audio/transcriptions?api-version=2024-10-21
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=2024-10-21
"---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 - Create
POST https://{endpoint}/openai/deployments/{deployment-id}/audio/translations?api-version=2024-10-21
Transcribes and translates input audio into English text.
URI Parameters
Required
Description
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
string
Deployment ID of the whisper model which was deployed.

For information about supported models, see [/azure/ai-foundry/openai/concepts/models#audio-models].
api-version
query
string
API version
Request Header
Required
Description
prompt
string
An optional text to guide the model's style or continue a previous audio segment. The prompt should be in English.
response_format
audioResponseFormat
Defines the format of the output.
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.
Responses
Status Code: 200
Description: OK
Content-Type
Description
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=2024-10-21
"---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=2024-10-21
"---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=2024-10-21
Generates a batch of images from a text caption on a given dall-e model deployment
URI Parameters
Required
Description
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
string
Deployment ID of the dall-e model which was deployed.
api-version
query
string
API version
Request Header
Required
Description
prompt
string
A text description of the desired image(s). The maximum length is 4,000 characters.
integer
The number of images to generate.
imageSize
The size of the generated images.
1024x1024
response_format
imagesResponseFormat
The format in which the generated images are returned.
string
A unique identifier representing your end-user, which can help to monitor and detect abuse.
quality
imageQuality
The quality of the image that will be generated.
standard
style
imageStyle
The style of the generated images.
vivid
Responses
Status Code: 200
Description: Ok
Content-Type
Description
Example
Creates images given a prompt.
POST https://{endpoint}/openai/deployments/{deployment-id}/images/generations?api-version=2024-10-21
 "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
Description
Required
Default
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.
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
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.
revised_prompt
string
The prompt that was used to generate the image, if there was any revision to the prompt.
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.
Description
Required
Default
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.
Description
Required
Default
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.
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 properties defined for this component.
dalleContentFilterResults
Information about the content filtering results.
Description
Required
Default
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.
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.
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.
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.
False
string or array
Up to four sequences where the API will stop generating further tokens.
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). This value is now deprecated in favor of max_completion_tokens, and isn't compatible with o1 series models.
max_completion_tokens
integer
An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.
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.
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.
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.
string
A unique identifier representing your end-user, which can help Azure OpenAI to monitor and detect abuse.
prompt
string or array
The prompt(s) to generate completions for, encoded as a string, array of strings, array of tokens, or array of token arrays.

Note that <|endoftext|> is the document separator that the model sees during training, so if a prompt isn't specified the model will generate as if from the beginning of a new document.

best_of
integer
Generates best_of completions server-side and returns the "best" (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.

boolean
Echo back the prompt in addition to the completion

False
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.

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 GPT 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.

As an example, you can pass {"50256": -100} to prevent the <|endoftext|> token from being generated.

logprobs
integer
Include the log probabilities on the logprobs most likely output tokens, as well the chosen tokens. For example, if logprobs is 5, the API will return a list of the five 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.

The maximum value for logprobs is 5.

max_tokens
integer
The maximum number of tokens that can be generated in the completion.

The token count of your prompt plus max_tokens can't exceed the model's context length.
integer
How many completions to generate for each prompt.

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.

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.

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't guaranteed, and you should refer to the system_fingerprint response parameter to monitor changes in the backend.

string or array
Up to four sequences where the API will stop generating further tokens. The returned text won't contain the stop sequence.

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.
False
suffix
string
The suffix that comes after a completion of inserted text.

This parameter is only supported for gpt-3.5-turbo-instruct.

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.

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.

string
A unique identifier representing your end-user, which can help to monitor and detect abuse.

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.
system_fingerprint
string
This fingerprint represents the backend configuration that the model runs with.

Can be used in conjunction with the seed request parameter to understand when backend changes have been made that might impact determinism.

object
The object type, which is always "text_completion"
Possible values: text_completion
usage
completionUsage
Usage statistics for the completion request.
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.

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.

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.
False
string or array
Up to four sequences where the API will stop generating further tokens.

max_tokens
integer
The maximum number of tokens that can be generated in the chat completion.

The total length of input tokens and generated tokens is limited by the model's context length.
max_completion_tokens
integer
An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.
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.

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.

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.

string
A unique identifier representing your end-user, which can help to monitor and detect abuse.

messages
array
A list of messages comprising the conversation so far.
data_sources
array
The configuration entries for Azure OpenAI chat extensions that use them.
  This additional specification is only compatible with Azure OpenAI.
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.
False
top_logprobs
integer
An integer between 0 and 20 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.
integer
How many chat completion choices to generate for each input message. Note that you'll be charged based on the number of generated tokens across all of the choices. Keep n as 1 to minimize costs.
parallel_tool_calls
ParallelToolCalls
Whether to enable parallel function calling during tool use.
response_format
ResponseFormatText or ResponseFormatJsonObject or ResponseFormatJsonSchema
An object specifying the format that the model must output. Compatible with GPT-4o, GPT-4o mini, GPT-4 Turbo and all GPT-3.5 Turbo models newer than gpt-3.5-turbo-1106.

Setting to { "type": "json_schema", "json_schema": {...} } enables Structured Outputs which guarantees the model will match your supplied JSON schema.

Setting to { "type": "json_object" } enables JSON mode, which guarantees the message the model generates is valid JSON.

Important: when using JSON mode, you must also instruct the model to produce JSON yourself via a system or user message. Without this, the model may generate an unending stream of whitespace until the generation reaches the token limit, resulting in a long-running and seemingly "stuck" request. Also note that the message content may be partially cut off if finish_reason="length", which indicates the generation exceeded max_tokens or the conversation exceeded the max context length.

integer
This feature is in Beta.
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't guaranteed, and you should refer to the system_fingerprint response parameter to monitor changes in the backend.

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. A max of 128 functions are supported.

tool_choice
chatCompletionToolChoiceOption
Controls which (if any) tool is called by the model. none means the model won't call any tool and instead generates a message. auto means the model can pick between generating a message or calling one or more tools. required means the model must call one or more tools. Specifying a particular tool via {"type": "function", "function": {"name": "my_function"}} forces the model to call that tool. none is the default when no tools are present. auto is the default if tools are present.
function_call
string or chatCompletionFunctionCallOption
Deprecated in favor of tool_choice.

Controls which (if any) function is called by the model.
none means the model won't 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 {"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.

functions
array
Deprecated in favor of tools.

A list of functions the model may generate JSON inputs for.

description
string
A description of what the function does, used by the model to choose when and how to call the function.
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
FunctionParameters
The parameters the functions accepts, described as a JSON Schema object. See the guide for examples, and the JSON Schema reference for documentation about the format. 

Omitting parameters defines a function with an empty parameter list.
string
An optional name for the participant. Provides the model information to differentiate between participants of the same role.
string
An optional name for the participant. Provides the model information to differentiate between participants of the same role.
content
string or array
The contents of the assistant message. Required unless tool_calls or function_call is specified.

refusal
string
The refusal message by the assistant.
The role of the messages author, in this case assistant.
Possible values: assistant
string
An optional name for the participant. Provides the model information to differentiate between participants of the same role.
tool_calls
chatCompletionMessageToolCalls
The tool calls generated by the model, such as function calls.
function_call
object
Deprecated and replaced by tool_calls. The name and arguments of a function that should be called, as generated by the model.
The role of the messages author, in this case function.
Possible values: function
content
string
The contents of the function message.
string
The name of the function to call.
chatCompletionRequestSystemMessageContentPart
This component can be one of the following:




    

chatCompletionRequestUserMessageContentPart
This component can be one of the following:
chatCompletionRequestAssistantMessageContentPart
This component can be one of the following:
chatCompletionRequestToolMessageContentPart
This component can be one of the following:
chatCompletionRequestMessageContentPartText
Description
Required
Default
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.
Description
Required
Default
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.
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.
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
azureSearchChatExtensionConfiguration
A specific representation of configurable options for Azure Search when using it as an Azure OpenAI chat
extension.
Description
Required
Default
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.
parameters
azureSearchChatExtensionParameters
Parameters for Azure Search when used as an Azure OpenAI chat extension.
authentication
onYourDataApiKeyAuthenticationOptions or onYourDataSystemAssignedManagedIdentityAuthenticationOptions or onYourDataUserAssignedManagedIdentityAuthenticationOptions
top_n_documents
integer
The configured top number of documents to feature for the configured query.
in_scope
boolean
Whether queries should be restricted to use of indexed data.
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.
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.
endpoint
string
The absolute endpoint path for the Azure Search resource to use.
index_name
string
The name of the index to use as available in the referenced Azure Search resource.
fields_mapping
azureSearchIndexFieldMappingOptions
Optional settings to control how fields are processed when using a configured Azure Search resource.
query_type
azureSearchQueryType
The type of Azure Search retrieval query that should be executed when using it as an Azure OpenAI chat extension.
semantic_configuration
string
The additional semantic configuration for the query.
filter
string
Search filter.
embedding_dependency
onYourDataEndpointVectorizationSource or onYourDataDeploymentNameVectorizationSource
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
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.
Description
Required
Default
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.
parameters
azureCosmosDBChatExtensionParameters
Parameters to use when configuring Azure OpenAI On Your Data chat extensions when using Azure Cosmos DB for
MongoDB vCore.
azureCosmosDBChatExtensionParameters
Parameters to use when configuring Azure OpenAI On Your Data chat extensions when using Azure Cosmos DB for
MongoDB vCore.
Description
Required
Default
authentication
onYourDataConnectionStringAuthenticationOptions
The authentication options for Azure OpenAI On Your Data when using a connection string.
top_n_documents
integer
The configured top number of documents to feature for the configured query.
in_scope
boolean
Whether queries should be restricted to use of indexed data.
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.
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.
database_name
string
The MongoDB vCore database name to use with Azure Cosmos DB.
container_name
string
The name of the Azure Cosmos DB resource container.
index_name
string
The MongoDB vCore index name to use with Azure Cosmos DB.
fields_mapping
azureCosmosDBFieldMappingOptions
Optional settings to control how fields are processed when using a configured Azure Cosmos DB resource.
embedding_dependency
onYourDataEndpointVectorizationSource or onYourDataDeploymentNameVectorizationSource
onYourDataAuthenticationType
The authentication types supported with Azure OpenAI On Your Data.
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
onYourDataAuthenticationType
The authentication types supported with Azure OpenAI On Your Data.
string
The API key to use for authentication.
onYourDataAuthenticationType
The authentication types supported with Azure OpenAI On Your Data.
connection_string
string
The connection string to use for authentication.
onYourDataAuthenticationType
The authentication types supported with Azure OpenAI On Your Data.
onYourDataAuthenticationType
The authentication types supported with Azure OpenAI On Your Data.
managed_identity_resource_id
string
The resource ID of the user-assigned managed identity to use for authentication.
onYourDataVectorizationSourceType
Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with
vector search.
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
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.
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.
Description
Required
Default
onYourDataVectorizationSourceType
Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with
vector search.
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.
onYourDataEndpointVectorizationSource
The details of a vectorization source, used by Azure OpenAI On Your Data when applying vector search, that is based
on a public Azure OpenAI endpoint call for embeddings.
Description
Required
Default
onYourDataVectorizationSourceType
Represents the available sources Azure OpenAI On Your Data can use to configure vectorization of data for use with
vector search.
authentication
onYourDataApiKeyAuthenticationOptions
The authentication options for Azure OpenAI On Your Data when using an API key.
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.
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.
Description
Required
Default
citations
array
The data source retrieval result, used to generate the assistant message in the response.
intent
string
The detected intent from the chat history, used to pass to the next turn to carry over the context.
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
The role of the messages author, in this case function.
Possible values: function
string
The contents of the message.
content
string
The contents of the message.
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.
choices
array
A list of chat completion choices. Can be more than one if n is greater than 1.
created
integer
The Unix timestamp (in seconds) of when the chat completion was created.
model
string
The model used for the chat completion.
system_fingerprint
string
This fingerprint represents the backend configuration that the model runs with.

Can be used in conjunction with the seed request parameter to understand when backend changes have been made that might impact determinism.

object
The object type, which is always chat.completion.
Possible values: chat.completion
usage
completionUsage
Usage statistics for the completion request.
choices
array
A list of chat completion choices. Can contain more than one elements if n is greater than 1.

created
integer
The Unix timestamp (in seconds) of when the chat completion was created. Each chunk has the same timestamp.
model
string
The model to generate the completion.
system_fingerprint
string
This fingerprint represents the backend configuration that the model runs with.
Can be used in conjunction with the seed request parameter to understand when backend changes have been made that might impact determinism.

object
The object type, which is always chat.completion.chunk.
Possible values: chat.completion.chunk
function_call
object
Deprecated and replaced by tool_calls. The name and arguments of a function that should be called, as generated by the model.
tool_calls
array
The role of the author of this message.
Possible values: system, user, assistant, tool
refusal
string
The refusal message generated by the model.
The type of the tool. Currently, only function is supported.
Possible values: function
function
object
include_usage
boolean
If set, an additional chunk will be streamed before the data: [DONE] message. The usage field on this chunk shows the token usage statistics for the entire request, and the choices field will always be an empty array. All other chunks will also include a usage field, but with a null value.

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's no bytes representation for the token.
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.
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.
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.
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) tool is called by the model. none means the model won't call any tool and instead generates a message. auto means the model can pick between generating a message or calling one or more tools. required means the model must call one or more tools. Specifying a particular tool via {"type": "function", "function": {"name": "my_function"}} forces the model to call that tool. none is the default when no tools are present. auto is the default if tools are present.
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.
Description
Required
Default
The type of the tool. Currently, only function is supported.
Possible values: function
function
object
ParallelToolCalls
Whether to enable parallel function calling during tool use.
No properties defined for this component.
chatCompletionMessageToolCalls
The tool calls generated by the model, such as function calls.
No properties defined for this component.
chatCompletionFunctionCall
Deprecated and replaced by tool_calls. The name and arguments of a function that should be called, as generated by the model.
Description
Required
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 generate parameters not defined by your function schema. Validate the arguments in your code before calling your function.
The type of the tool. Currently, only function is supported.
Possible values: function
function
FunctionObject
FunctionParameters
The parameters the functions accepts, described as a JSON Schema object. See the guide for examples, and the JSON Schema reference for documentation about the format.
Omitting parameters defines a function with an empty parameter list.
No properties defined for this component.
FunctionObject
Description
Required
Default
description
string
A description of what the function does, used by the model to choose when and how to call the function.
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
FunctionParameters
The parameters the functions accepts, described as a JSON Schema object. See the guide for examples, and the JSON Schema reference for documentation about the format. 

Omitting parameters defines a function with an empty parameter list.
strict
boolean
Whether to enable strict schema adherence when generating the function call. If set to true, the model will follow the exact schema defined in the parameters field. Only a subset of JSON Schema is supported when strict is true.
False
ResponseFormatText
Description
Required
Default
The type of response format being defined: json_object
Possible values: json_object
ResponseFormatJsonSchemaSchema
The schema for the response format, described as a JSON Schema object.
No properties defined for this component.
ResponseFormatJsonSchema
Description
Required
Default
The type of response format being defined: json_schema
Possible values: json_schema
json_schema
object
strict
boolean
Whether to enable strict schema adherence when generating the output. If set to true, the model will always follow the exact schema defined in the schema field. Only a subset of JSON Schema is supported when strict is true.
False
chatCompletionChoiceCommon
Description
Required
Default
prompt
string
An optional text to guide the model's style or continue a previous audio segment. The prompt should be in English.
response_format
audioResponseFormat
Defines the format of the output.
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.
audioResponse
Translation or transcription response when response_format was json
Description
Required
Default
audioResponseFormat
Defines the format of the output.
Description: Defines the format of the output.
Type: string
Default:
Enum Values:
verbose_json
createTranscriptionRequest
Transcription request.
Description
Required
Default
prompt
string
An optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.
response_format
audioResponseFormat
Defines the format of the output.
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.
language
string
The language of the input audio. Supplying the input language in ISO-639-1 format will improve accuracy and latency.
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
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
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
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
prompt
string
A text description of the desired image(s). The maximum length is 4,000 characters.
integer
The number of images to generate.
imageSize
The size of the generated images.
1024x1024
response_format
imagesResponseFormat
The format in which the generated images are returned.
string
A unique identifier representing your end-user, which can help to monitor and detect abuse.
quality
imageQuality
The quality of the image that will be generated.
standard
style
imageStyle
The style of the generated images.
vivid
generateImagesResponse
Description
Required
Default
content_filter_results
dalleContentFilterResults
Information about the content filtering results.
revised_prompt
string
The prompt that was used to generate the image, if there was any revision to the prompt.
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.
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 (Preview)
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.