Azure OpenAI fine-tuning

API Version: v1

Server: {endpoint}/openai/v1 — Azure AI Foundry Models APIs

Server Variables:

Variable	Default	Description
endpoint		A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).

Authentication

ApiKeyAuth (API Key)

Pass your API key in the api-key header.

ApiKeyAuth_ (API Key)

Pass your API key in the authorization header.

OAuth2Auth (OAuth 2.0)

Flow: implicit

Authorization URL: https://login.microsoftonline.com/common/oauth2/v2.0/authorize

Scopes:

• https://cognitiveservices.azure.com/.default

Security Requirements

Endpoints accept any one of the following authentication methods:

1. ApiKeyAuth
2. ApiKeyAuth_
3. OAuth2Auth (scopes: https://cognitiveservices.azure.com/.default)

Run grader

POST {endpoint}/openai/v1/fine_tuning/alpha/graders/run

Run a grader.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
grader	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderPython or OpenAI.GraderScoreModel or OpenAI.GraderMulti or GraderEndpoint	The grader used for the fine-tuning job.	Yes
item	OpenAI.RunGraderRequestItem		No
model_sample	string	The model sample to be evaluated. This value will be used to populate the sample namespace. The output_json variable will be populated if the model sample is a valid JSON string.	Yes

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.RunGraderResponse

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Validate grader

POST {endpoint}/openai/v1/fine_tuning/alpha/graders/validate

Validate a grader.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
grader	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderPython or OpenAI.GraderScoreModel or OpenAI.GraderMulti or GraderEndpoint		No

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ValidateGraderResponse

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

List fine tuning checkpoint permissions

GET {endpoint}/openai/v1/fine_tuning/checkpoints/{fine_tuned_model_checkpoint}/permissions

List checkpoint permissions

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuned_model_checkpoint	path	Yes	string	The ID of the fine-tuned model checkpoint to get permissions for.
project_id	query	No	string	The ID of the project to get permissions for. Serialization: explode: false
after	query	No	string	Identifier for the last permission ID from the previous pagination request. Serialization: explode: false
limit	query	No	integer (int32)	Number of permissions to retrieve. Serialization: explode: false
order	query	No	string Possible values: ascending, descending	The order in which to retrieve permissions. Serialization: explode: false

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ListFineTuningCheckpointPermissionResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Create fine tuning checkpoint permission

POST {endpoint}/openai/v1/fine_tuning/checkpoints/{fine_tuned_model_checkpoint}/permissions

Create checkpoint permissions

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuned_model_checkpoint	path	Yes	string	The ID of the fine-tuned model checkpoint to create a permission for.

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
project_ids	array of string	The project identifiers to grant access to.	Yes

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ListFineTuningCheckpointPermissionResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Delete fine tuning checkpoint permission

DELETE {endpoint}/openai/v1/fine_tuning/checkpoints/{fine_tuned_model_checkpoint}/permissions/{permission_id}

Delete checkpoint permission

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuned_model_checkpoint	path	Yes	string	The ID of the fine-tuned model checkpoint to delete a permission for.
permission_id	path	Yes	string	The ID of the fine-tuned model checkpoint permission to delete.

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.DeleteFineTuningCheckpointPermissionResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Create fine tuning job

POST {endpoint}/openai/v1/fine_tuning/jobs

Creates a fine-tuning job which begins the process of creating a new model from a given dataset.

Response includes details of the enqueued job including job status and the name of the fine-tuned models once complete.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
hyperparameters	OpenAI.CreateFineTuningJobRequestHyperparameters (deprecated)		No
└─ batch_size	string or integer		No	auto
└─ learning_rate_multiplier	string or number		No	auto
└─ n_epochs	string or integer		No	auto
integrations	array of OpenAI.CreateFineTuningJobRequestIntegrations or null	A list of integrations to enable for your fine-tuning job.	No
metadata	OpenAI.Metadata or null		No
method	OpenAI.FineTuneMethod	The method used for fine-tuning.	No
model	string (see valid models below)	The name of the model to fine-tune. You can select one of the supported models.	Yes
seed	integer or null	The seed controls the reproducibility of the job. Passing in the same seed and job parameters should produce the same results, but may differ in rare cases. If a seed is not specified, one will be generated for you. Constraints: min: 0, max: 2147483647	No
suffix	string or null	A string of up to 64 characters that will be added to your fine-tuned model name. For example, a suffix of "custom-model-name" would produce a model name like ft:gpt-4o-mini:openai:custom-model-name:7p4lURel. Constraints: minLength: 1, maxLength: 64	No
training_file	string	The ID of an uploaded file that contains training data. Your dataset must be formatted as a JSONL file. Additionally, you must upload your file with the purpose fine-tune. The contents of the file should differ depending on if the model uses the chat, completions format, or if the fine-tuning method uses the preference format.	Yes
validation_file	string or null	The ID of an uploaded file that contains validation data. If you provide this file, the data is used to generate validation metrics periodically during fine-tuning. These metrics can be viewed in the fine-tuning results file. The same data should not be present in both train and validation files. Your dataset must be formatted as a JSONL file. You must upload your file with the purpose fine-tune.	No

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.FineTuningJob

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

List paginated fine tuning jobs

GET {endpoint}/openai/v1/fine_tuning/jobs

List your organization's fine-tuning jobs

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
after	query	No	string	Identifier for the last job from the previous pagination request. Serialization: explode: false
limit	query	No	integer (int32)	Number of fine-tuning jobs to retrieve. Serialization: explode: false

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ListPaginatedFineTuningJobsResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Retrieve fine tuning job

GET {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}

Get info about a fine-tuning job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job.

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.FineTuningJob

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Cancel fine tuning job

POST {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/cancel

Immediately cancel a fine-tune job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job to cancel.

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.FineTuningJob

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

List fine tuning job checkpoints

GET {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/checkpoints

List the checkpoints for a fine-tuning job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job to get checkpoints for.
after	query	No	string	Identifier for the last checkpoint ID from the previous pagination request. Serialization: explode: false
limit	query	No	integer (int32)	Number of checkpoints to retrieve. Serialization: explode: false

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ListFineTuningJobCheckpointsResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Fine tuning - copy checkpoint

POST {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/checkpoints/{fine_tuning_checkpoint_id}/copy

Creates a copy of a fine-tuning checkpoint at the given destination account and region.

NOTE: This Azure OpenAI API is in preview and subject to change.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string
fine_tuning_checkpoint_id	path	Yes	string

Request Header

Name	Required	Type	Description
aoai-copy-ft-checkpoints	Yes	string Possible values: preview	Enables access to checkpoint copy operations for models, an AOAI preview feature. This feature requires the 'aoai-copy-ft-checkpoints' header to be set to 'preview'.
accept	Yes	string Possible values: application/json

Request Body

Content-Type: application/json

Name	Type	Description	Required	Default
destinationResourceId	string	The ID of the destination Resource to copy.	Yes
region	string	The region to copy the model to.	Yes

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	CopyModelResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Fine tuning - get checkpoint

GET {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/checkpoints/{fine_tuning_checkpoint_id}/copy

Gets the status of a fine-tuning checkpoint copy.

NOTE: This Azure OpenAI API is in preview and subject to change.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string
fine_tuning_checkpoint_id	path	Yes	string

Request Header

Name	Required	Type	Description
aoai-copy-ft-checkpoints	Yes	string Possible values: preview	Enables access to checkpoint copy operations for models, an AOAI preview feature. This feature requires the 'aoai-copy-ft-checkpoints' header to be set to 'preview'.
accept	Yes	string Possible values: application/json

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	CopyModelResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

List fine tuning events

GET {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/events

Get status updates for a fine-tuning job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job to get events for.
after	query	No	string	Identifier for the last event from the previous pagination request. Serialization: explode: false
limit	query	No	integer (int32)	Number of events to retrieve. Serialization: explode: false

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.ListFineTuningJobEventsResponse

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Pause fine tuning job

POST {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/pause

Pause a fine-tune job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job to pause.

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.FineTuningJob

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Resume fine tuning job

POST {endpoint}/openai/v1/fine_tuning/jobs/{fine_tuning_job_id}/resume

Resume a paused fine-tune job.

URI Parameters

Name	In	Required	Type	Description
endpoint	server	Yes	string	A supported Azure AI Foundry Models APIs endpoint, including protocol and hostname. For example: https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/).
api-version	query	No	string Possible values: v1, preview	The explicit Azure AI Foundry Models API version to use for this request. v1 if not otherwise specified.
fine_tuning_job_id	path	Yes	string	The ID of the fine-tuning job to resume.

Responses

Status Code: 200

Description: The request has succeeded.

Content-Type	Type	Description
application/json	OpenAI.FineTuningJob

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Status Code: default

Description: An unexpected error response.

Content-Type	Type	Description
application/json	object

Name	Type	Description	Required	Default
code	string or null		Yes
inner_error			No
message	string		Yes
param	string or null		No
type	string		No

Response Headers:

Header	Type	Description
apim-request-id	string	A request ID used for troubleshooting purposes.

Components

AzureAIFoundryModelsApiVersion

Property	Value
Type	string
Values	v1 preview

AzureFineTuneReinforcementMethod

Name	Type	Description	Required	Default
grader	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderScoreModel or OpenAI.GraderMulti or GraderEndpoint		Yes
hyperparameters	OpenAI.FineTuneReinforcementHyperparameters	The hyperparameters used for the reinforcement fine-tuning job.	No
response_format	ResponseFormatJSONSchemaRequest		No
└─ json_schema	object	JSON Schema for the response format	Yes
└─ type	enum	Type of response format Possible values: json_schema	Yes

CopiedAccountDetails

Name	Type	Description	Required	Default
destinationResourceId	string	The ID of the destination resource where the model was copied to.	Yes
region	string	The region where the model was copied to.	Yes
status	enum	The status of the copy operation. Possible values: Completed, Failed, InProgress	Yes

CopyModelRequest

Name	Type	Description	Required	Default
destinationResourceId	string	The ID of the destination Resource to copy.	Yes
region	string	The region to copy the model to.	Yes

CopyModelResponse

Name	Type	Description	Required	Default
checkpointedModelName	string	The ID of the copied model.	Yes
copiedAccountDetails	array of CopiedAccountDetails	The ID of the destination resource id where it was copied	Yes
fineTuningJobId	string	The ID of the fine-tuning job that the checkpoint was copied from.	Yes

GraderEndpoint

Endpoint grader configuration for external HTTP endpoint evaluation

Name	Type	Description	Required	Default
headers	object or null	Optional HTTP headers to include in requests to the endpoint	No
name	string	The name of the grader	Yes
pass_threshold	number or null	Optional threshold score above which the grade is considered passing If not specified, all scores are considered valid	No
rate_limit	integer or null	Optional rate limit for requests per second to the endpoint Must be a positive integer	No
type	enum	Possible values: endpoint	Yes	endpoint
url	string	The HTTPS URL of the endpoint to call for grading Constraints: pattern: ^https://	Yes

OpenAI.CreateFineTuningCheckpointPermissionRequest

Name	Type	Description	Required	Default
project_ids	array of string	The project identifiers to grant access to.	Yes

OpenAI.CreateFineTuningJobRequest

Valid models:

babbage-002
davinci-002
gpt-3.5-turbo
gpt-4o-mini

Name	Type	Description	Required	Default
hyperparameters	OpenAI.CreateFineTuningJobRequestHyperparameters (deprecated)		No
└─ batch_size	string or integer		No	auto
└─ learning_rate_multiplier	string or number		No	auto
└─ n_epochs	string or integer		No	auto
integrations	array of OpenAI.CreateFineTuningJobRequestIntegrations or null	A list of integrations to enable for your fine-tuning job.	No
metadata	OpenAI.Metadata or null		No
method	OpenAI.FineTuneMethod	The method used for fine-tuning.	No
model	string (see valid models below)	The name of the model to fine-tune. You can select one of the supported models.	Yes
seed	integer or null	The seed controls the reproducibility of the job. Passing in the same seed and job parameters should produce the same results, but may differ in rare cases. If a seed is not specified, one will be generated for you. Constraints: min: 0, max: 2147483647	No
suffix	string or null	A string of up to 64 characters that will be added to your fine-tuned model name. For example, a suffix of "custom-model-name" would produce a model name like ft:gpt-4o-mini:openai:custom-model-name:7p4lURel. Constraints: minLength: 1, maxLength: 64	No
training_file	string	The ID of an uploaded file that contains training data. Your dataset must be formatted as a JSONL file. Additionally, you must upload your file with the purpose fine-tune. The contents of the file should differ depending on if the model uses the chat, completions format, or if the fine-tuning method uses the preference format.	Yes
validation_file	string or null	The ID of an uploaded file that contains validation data. If you provide this file, the data is used to generate validation metrics periodically during fine-tuning. These metrics can be viewed in the fine-tuning results file. The same data should not be present in both train and validation files. Your dataset must be formatted as a JSONL file. You must upload your file with the purpose fine-tune.	No

OpenAI.CreateFineTuningJobRequestHyperparameters

Name	Type	Description	Required	Default
batch_size	string or integer		No
learning_rate_multiplier	string or number		No
n_epochs	string or integer		No

OpenAI.CreateFineTuningJobRequestIntegrations

Name	Type	Description	Required	Default
type	enum	Possible values: wandb	Yes
wandb	OpenAI.CreateFineTuningJobRequestIntegrationsWandb		Yes

OpenAI.CreateFineTuningJobRequestIntegrationsWandb

Name	Type	Description	Required	Default
entity	string or null		No
name	string or null		No
project	string		Yes
tags	array of string		No

OpenAI.DeleteFineTuningCheckpointPermissionResponse

Name	Type	Description	Required	Default
deleted	boolean	Whether the fine-tuned model checkpoint permission was successfully deleted.	Yes
id	string	The ID of the fine-tuned model checkpoint permission that was deleted.	Yes
object	enum	The object type, which is always "checkpoint.permission". Possible values: checkpoint.permission	Yes

OpenAI.EvalGraderScoreModelSamplingParams

Name	Type	Description	Required	Default
max_completions_tokens	integer or null		No
reasoning_effort	OpenAI.ReasoningEffort	Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response. - gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1. - All models before gpt-5.1 default to medium reasoning effort, and do not support none. - The gpt-5-pro model defaults to (and only supports) high reasoning effort. - xhigh is supported for all models after gpt-5.1-codex-max.	No
seed	integer or null		No
temperature	number or null		No
top_p	number or null		No

OpenAI.EvalItem

Eval message object

A message input to the model with a role indicating instruction following hierarchy. Instructions given with the developer or system role take precedence over instructions given with the user role. Messages with the assistant role are presumed to have been generated by the model in previous interactions.

Name	Type	Description	Required	Default
content	OpenAI.EvalItemContent	Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.	Yes
role	enum	The role of the message input. One of user, assistant, system, or developer. Possible values: user, assistant, system, developer	Yes
type	enum	The type of the message input. Always message. Possible values: message	No

OpenAI.EvalItemContent

Eval content

Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.

This component can be one of the following:

• OpenAI.EvalItemContentItem
• OpenAI.EvalItemContentArray

OpenAI.EvalItemContentArray

An array of Input text, Output text, Input image, and Input audio

A list of inputs, each of which may be either an input text, output text, input image, or input audio object.

Array of: OpenAI.EvalItemContentItem

OpenAI.EvalItemContentItem

Eval content item

A single content item: input text, output text, input image, or input audio.

This component can be one of the following:

• OpenAI.EvalItemContentText
• OpenAI.EvalItemContentItemObject

OpenAI.EvalItemContentItemObject

Eval content item

A single content item: input text, output text, input image, or input audio.

Discriminator for OpenAI.EvalItemContentItemObject

This component uses the property type to discriminate between different types:

Type Value	Schema
output_text	OpenAI.EvalItemContentOutputText
input_image	OpenAI.EvalItemInputImage
input_audio	OpenAI.InputAudio
input_text	OpenAI.EvalItemContentItemObjectInputTextContent

Name	Type	Description	Required	Default
type	OpenAI.EvalItemContentItemObjectType		Yes

OpenAI.EvalItemContentItemObjectInputTextContent

Input text

A text input to the model.

Name	Type	Description	Required	Default
text	string	The text input to the model.	Yes
type	enum	The type of the input item. Always input_text. Possible values: input_text	Yes	input_text

OpenAI.EvalItemContentItemObjectType

Property	Value
Type	string
Values	input_text output_text input_image input_audio

OpenAI.EvalItemContentOutputText

Output text

A text output from the model.

Name	Type	Description	Required	Default
text	string	The text output from the model.	Yes
type	enum	The type of the output text. Always output_text. Possible values: output_text	Yes

OpenAI.EvalItemContentText

Text input

A text input to the model.

Type: string

OpenAI.EvalItemInputImage

Input image

An image input block used within EvalItem content arrays.

Name	Type	Description	Required	Default
detail	string	The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.	No
image_url	string (uri)	The URL of the image input.	Yes
type	enum	The type of the image input. Always input_image. Possible values: input_image	Yes

OpenAI.FineTuneDPOHyperparameters

The hyperparameters used for the DPO fine-tuning job.

Name	Type	Description	Required	Default
batch_size	string or integer	Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.	No
beta	string or number	The beta value for the DPO method. A higher beta value will increase the weight of the penalty between the policy and reference model.	No
learning_rate_multiplier	string or number	Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.	No
n_epochs	string or integer	The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.	No

OpenAI.FineTuneDPOMethod

Configuration for the DPO fine-tuning method.

Name	Type	Description	Required	Default
hyperparameters	OpenAI.FineTuneDPOHyperparameters	The hyperparameters used for the DPO fine-tuning job.	No

OpenAI.FineTuneMethod

The method used for fine-tuning.

Name	Type	Description	Required	Default
dpo	OpenAI.FineTuneDPOMethod	Configuration for the DPO fine-tuning method.	No
reinforcement	AzureFineTuneReinforcementMethod		No
supervised	OpenAI.FineTuneSupervisedMethod	Configuration for the supervised fine-tuning method.	No
type	enum	The type of method. Is either supervised, dpo, or reinforcement. Possible values: supervised, dpo, reinforcement	Yes

OpenAI.FineTuneReinforcementHyperparameters

The hyperparameters used for the reinforcement fine-tuning job.

Name	Type	Description	Required	Default
batch_size	string or integer	Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.	No
compute_multiplier	string or number	Multiplier on amount of compute used for exploring search space during training.	No
eval_interval	string or integer	The number of training steps between evaluation runs.	No
eval_samples	string or integer	Number of evaluation samples to generate per training step.	No
learning_rate_multiplier	string or number	Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.	No
n_epochs	string or integer	The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.	No
reasoning_effort	enum	Level of reasoning effort. Possible values: default, low, medium, high	No	default

OpenAI.FineTuneSupervisedHyperparameters

The hyperparameters used for the fine-tuning job.

Name	Type	Description	Required	Default
batch_size	string or integer	Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.	No
learning_rate_multiplier	string or number	Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.	No
n_epochs	string or integer	The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.	No

OpenAI.FineTuneSupervisedMethod

Configuration for the supervised fine-tuning method.

Name	Type	Description	Required	Default
hyperparameters	OpenAI.FineTuneSupervisedHyperparameters	The hyperparameters used for the fine-tuning job.	No

OpenAI.FineTuningCheckpointPermission

FineTuningCheckpointPermission

The checkpoint.permission object represents a permission for a fine-tuned model checkpoint.

Name	Type	Description	Required	Default
created_at	integer (unixtime)	The Unix timestamp (in seconds) for when the permission was created.	Yes
id	string	The permission identifier, which can be referenced in the API endpoints.	Yes
object	enum	The object type, which is always "checkpoint.permission". Possible values: checkpoint.permission	Yes
project_id	string	The project identifier that the permission is for.	Yes

OpenAI.FineTuningIntegration

Fine-Tuning Job Integration

Name	Type	Description	Required	Default
type	enum	The type of the integration being enabled for the fine-tuning job Possible values: wandb	Yes
wandb	OpenAI.FineTuningIntegrationWandb		Yes
└─ entity	string or null		No
└─ name	string or null		No
└─ project	string		Yes
└─ tags	array of string		No

OpenAI.FineTuningIntegrationWandb

Name	Type	Description	Required	Default
entity	string or null		No
name	string or null		No
project	string		Yes
tags	array of string		No

OpenAI.FineTuningJob

FineTuningJob

The fine_tuning.job object represents a fine-tuning job that has been created through the API.

Name	Type	Description	Required	Default
created_at	integer (unixtime)	The Unix timestamp (in seconds) for when the fine-tuning job was created.	Yes
error	OpenAI.FineTuningJobError or null		Yes
estimated_finish	string or null		No
fine_tuned_model	string or null		Yes
finished_at	string or null		Yes
hyperparameters	OpenAI.FineTuningJobHyperparameters		Yes
└─ batch_size	string or integer or null		No	auto
└─ learning_rate_multiplier	string or number		No	auto
└─ n_epochs	string or integer		No	auto
id	string	The object identifier, which can be referenced in the API endpoints.	Yes
integrations	array of OpenAI.FineTuningIntegration or null		No
metadata	OpenAI.Metadata or null		No
method	OpenAI.FineTuneMethod	The method used for fine-tuning.	No
model	string	The base model that is being fine-tuned.	Yes
object	enum	The object type, which is always "fine_tuning.job". Possible values: fine_tuning.job	Yes
organization_id	string	The organization that owns the fine-tuning job.	Yes
result_files	array of string	The compiled results file ID(s) for the fine-tuning job. You can retrieve the results with the Files API.	Yes
seed	integer	The seed used for the fine-tuning job.	Yes
status	enum	The current status of the fine-tuning job, which can be either validating_files, queued, running, succeeded, failed, or cancelled. Possible values: validating_files, queued, running, succeeded, failed, cancelled	Yes
trained_tokens	integer or null		Yes
training_file	string	The file ID used for training. You can retrieve the training data with the Files API.	Yes
validation_file	string or null		Yes

OpenAI.FineTuningJobCheckpoint

FineTuningJobCheckpoint

The fine_tuning.job.checkpoint object represents a model checkpoint for a fine-tuning job that is ready to use.

Name	Type	Description	Required	Default
created_at	integer (unixtime)	The Unix timestamp (in seconds) for when the checkpoint was created.	Yes
fine_tuned_model_checkpoint	string	The name of the fine-tuned checkpoint model that is created.	Yes
fine_tuning_job_id	string	The name of the fine-tuning job that this checkpoint was created from.	Yes
id	string	The checkpoint identifier, which can be referenced in the API endpoints.	Yes
metrics	OpenAI.FineTuningJobCheckpointMetrics		Yes
└─ full_valid_loss	number		No
└─ full_valid_mean_token_accuracy	number		No
└─ step	number		No
└─ train_loss	number		No
└─ train_mean_token_accuracy	number		No
└─ valid_loss	number		No
└─ valid_mean_token_accuracy	number		No
object	enum	The object type, which is always "fine_tuning.job.checkpoint". Possible values: fine_tuning.job.checkpoint	Yes
step_number	integer	The step number that the checkpoint was created at.	Yes

OpenAI.FineTuningJobCheckpointMetrics

Name	Type	Description	Required	Default
full_valid_loss	number		No
full_valid_mean_token_accuracy	number		No
step	number		No
train_loss	number		No
train_mean_token_accuracy	number		No
valid_loss	number		No
valid_mean_token_accuracy	number		No

OpenAI.FineTuningJobError

Name	Type	Description	Required	Default
code	string		Yes
message	string		Yes
param	string or null		Yes

OpenAI.FineTuningJobEvent

Fine-tuning job event object

Name	Type	Description	Required	Default
created_at	integer (unixtime)	The Unix timestamp (in seconds) for when the fine-tuning job was created.	Yes
data	OpenAI.FineTuningJobEventData		No
id	string	The object identifier.	Yes
level	enum	The log level of the event. Possible values: info, warn, error	Yes
message	string	The message of the event.	Yes
object	enum	The object type, which is always "fine_tuning.job.event". Possible values: fine_tuning.job.event	Yes
type	enum	The type of event. Possible values: message, metrics	No

OpenAI.FineTuningJobEventData

Type: object

OpenAI.FineTuningJobHyperparameters

Name	Type	Description	Required	Default
batch_size	string or integer or null		No
learning_rate_multiplier	string or number		No
n_epochs	string or integer		No

OpenAI.GraderMulti

MultiGrader

A MultiGrader object combines the output of multiple graders to produce a single score.

Name	Type	Description	Required	Default
calculate_output	string	A formula to calculate the output based on grader results.	Yes
graders	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderScoreModel or GraderEndpoint		Yes
name	string	The name of the grader.	Yes
type	enum	The object type, which is always multi. Possible values: multi	Yes	multi

OpenAI.GraderPython

PythonGrader

A PythonGrader object that runs a python script on the input.

Name	Type	Description	Required	Default
image_tag	string	The image tag to use for the python script.	No
name	string	The name of the grader.	Yes
source	string	The source code of the python script.	Yes
type	enum	The object type, which is always python. Possible values: python	Yes

OpenAI.GraderScoreModel

ScoreModelGrader

A ScoreModelGrader object that uses a model to assign a score to the input.

Name	Type	Description	Required	Default
input	array of OpenAI.EvalItem	The input messages evaluated by the grader. Supports text, output text, input image, and input audio content blocks, and may include template strings.	Yes
model	string	The model to use for the evaluation.	Yes
name	string	The name of the grader.	Yes
range	array of number	The range of the score. Defaults to [0, 1].	No
sampling_params	OpenAI.EvalGraderScoreModelSamplingParams		No
└─ max_completions_tokens	integer or null		No
└─ reasoning_effort	OpenAI.ReasoningEffort	Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response. - gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1. - All models before gpt-5.1 default to medium reasoning effort, and do not support none. - The gpt-5-pro model defaults to (and only supports) high reasoning effort. - xhigh is supported for all models after gpt-5.1-codex-max.	No
└─ seed	integer or null		No
└─ temperature	number or null		No
└─ top_p	number or null		No	1
type	enum	The object type, which is always score_model. Possible values: score_model	Yes

OpenAI.GraderStringCheck

StringCheckGrader

A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.

Name	Type	Description	Required	Default
input	string	The input text. This may include template strings.	Yes
name	string	The name of the grader.	Yes
operation	enum	The string check operation to perform. One of eq, ne, like, or ilike. Possible values: eq, ne, like, ilike	Yes
reference	string	The reference text. This may include template strings.	Yes
type	enum	The object type, which is always string_check. Possible values: string_check	Yes

OpenAI.GraderTextSimilarity

TextSimilarityGrader

A TextSimilarityGrader object which grades text based on similarity metrics.

Name	Type	Description	Required	Default
evaluation_metric	enum	The evaluation metric to use. One of cosine, fuzzy_match, bleu, gleu, meteor, rouge_1, rouge_2, rouge_3, rouge_4, rouge_5, or rouge_l. Possible values: cosine, fuzzy_match, bleu, gleu, meteor, rouge_1, rouge_2, rouge_3, rouge_4, rouge_5, rouge_l	Yes
input	string	The text being graded.	Yes
name	string	The name of the grader.	Yes
reference	string	The text being graded against.	Yes
type	enum	The type of grader. Possible values: text_similarity	Yes	text_similarity

OpenAI.InputAudio

Input audio

An audio input to the model.

Name	Type	Description	Required	Default
input_audio	OpenAI.InputAudioInputAudio		Yes
type	enum	The type of the input item. Always input_audio. Possible values: input_audio	Yes

OpenAI.InputAudioInputAudio

Name	Type	Description	Required	Default
data	string		Yes
format	enum	Possible values: mp3, wav	Yes

OpenAI.ListFineTuningCheckpointPermissionResponse

Name	Type	Description	Required	Default
data	array of OpenAI.FineTuningCheckpointPermission		Yes
first_id	string or null		No
has_more	boolean		Yes
last_id	string or null		No
object	enum	Possible values: list	Yes

OpenAI.ListFineTuningJobCheckpointsResponse

Name	Type	Description	Required	Default
data	array of OpenAI.FineTuningJobCheckpoint		Yes
first_id	string or null		No
has_more	boolean		Yes
last_id	string or null		No
object	enum	Possible values: list	Yes

OpenAI.ListFineTuningJobEventsResponse

Name	Type	Description	Required	Default
data	array of OpenAI.FineTuningJobEvent		Yes
has_more	boolean		Yes
object	enum	Possible values: list	Yes

OpenAI.ListPaginatedFineTuningJobsResponse

Name	Type	Description	Required	Default
data	array of OpenAI.FineTuningJob		Yes
has_more	boolean		Yes
object	enum	Possible values: list	Yes

OpenAI.Metadata

Set of 16 key-value pairs that can be attached to an object. This can be useful for storing additional information about the object in a structured format, and querying for objects via API or the dashboard. Keys are strings with a maximum length of 64 characters. Values are strings with a maximum length of 512 characters.

Type: object

OpenAI.ReasoningEffort

Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response.

• gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1.
• All models before gpt-5.1 default to medium reasoning effort, and do not support none.
• The gpt-5-pro model defaults to (and only supports) high reasoning effort.
• xhigh is supported for all models after gpt-5.1-codex-max.

Property	Value
Description	Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response. - gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1. - All models before gpt-5.1 default to medium reasoning effort, and do not support none. - The gpt-5-pro model defaults to (and only supports) high reasoning effort. - xhigh is supported for all models after gpt-5.1-codex-max.
Type	string
Values	none minimal low medium high xhigh

OpenAI.RunGraderRequest

RunGraderRequest

Name	Type	Description	Required	Default
grader	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderPython or OpenAI.GraderScoreModel or OpenAI.GraderMulti or GraderEndpoint	The grader used for the fine-tuning job.	Yes
item	OpenAI.RunGraderRequestItem		No
model_sample	string	The model sample to be evaluated. This value will be used to populate the sample namespace. The output_json variable will be populated if the model sample is a valid JSON string.	Yes

OpenAI.RunGraderRequestItem

Type: object

OpenAI.RunGraderResponse

Name	Type	Description	Required	Default
metadata	OpenAI.RunGraderResponseMetadata		Yes
model_grader_token_usage_per_model	object		Yes
reward	number		Yes
sub_rewards	object		Yes

OpenAI.RunGraderResponseMetadata

Name	Type	Description	Required	Default
errors	OpenAI.RunGraderResponseMetadataErrors		Yes
execution_time	number		Yes
name	string		Yes
sampled_model_name	string or null		Yes
scores	object		Yes
token_usage	integer or null		Yes
type	string		Yes

OpenAI.RunGraderResponseMetadataErrors

Name	Type	Description	Required	Default
formula_parse_error	boolean		Yes
invalid_variable_error	boolean		Yes
model_grader_parse_error	boolean		Yes
model_grader_refusal_error	boolean		Yes
model_grader_server_error	boolean		Yes
model_grader_server_error_details	string or null		Yes
other_error	boolean		Yes
python_grader_runtime_error	boolean		Yes
python_grader_runtime_error_details	string or null		Yes
python_grader_server_error	boolean		Yes
python_grader_server_error_type	string or null		Yes
sample_parse_error	boolean		Yes
truncated_observation_error	boolean		Yes
unresponsive_reward_error	boolean		Yes

OpenAI.ValidateGraderResponse

ValidateGraderResponse

Name	Type	Description	Required	Default
grader	OpenAI.GraderStringCheck or OpenAI.GraderTextSimilarity or OpenAI.GraderPython or OpenAI.GraderScoreModel or OpenAI.GraderMulti or GraderEndpoint	The grader used for the fine-tuning job.	No

ResponseFormatJSONSchemaRequest

Name	Type	Description	Required	Default
json_schema	object	JSON Schema for the response format	Yes
type	enum	Type of response format Possible values: json_schema	Yes