Model Version
A version of a registered model in the Workspace Model Registry. Each version points to the
model artifacts it was registered from and moves through the stages None, Staging,
Production and Archived.
ModelVersion object
A version of a registered model in the Workspace Model Registry. Each version points to the
model artifacts it was registered from and moves through the stages None, Staging,
Production and Archived.
- namestring
Name of the registered model that the version belongs to.
- versionstring
Version number of the model version, as a string. Version numbers start at 1 for each registered model.
- creation_timestampint64
Time when the model version was created, as a Unix timestamp in milliseconds.
- last_updated_timestampint64
Time when the metadata of the model version was last updated, as a Unix timestamp in milliseconds.
- user_idstring
User that created this
model_version.
- current_stagestring
Current stage of the model version. Valid values are
None,Staging,ProductionandArchived. A new model version starts inNone.
- descriptionstring
User-provided description of the model version.
- sourcestring
URI of the model artifacts that the model version was registered from, as given in the create request.
- run_idstring
ID of the MLflow run that produced the model artifacts, if
sourcecomes from a run in the MLflow tracking server.
- statusstring
Registration status of the model version. The version can be used once the status is
READY.PENDING_REGISTRATIONFAILED_REGISTRATIONREADY
- status_messagestring
Details on the current
status, such as the reason why registration failed.
- tagsarray of object
Key-value tags attached to the model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
URL of the MLflow run that generated the model version, as given in the create request. It is typically set when the run is hosted on a different tracking server than the registry.
Get a model version GA
GET
Get a model version.
API scopes: mlflow
Parameters
- namestringRequiredquery
Name of the registered model
- versionstringRequiredquery
Model version number
Response
- model_versionobject
Show child attributesHide child attributes
- namestring
Name of the registered model that the version belongs to.
- versionstring
Version number of the model version, as a string. Version numbers start at 1 for each registered model.
- creation_timestampint64
Time when the model version was created, as a Unix timestamp in milliseconds.
- last_updated_timestampint64
Time when the metadata of the model version was last updated, as a Unix timestamp in milliseconds.
- user_idstring
User that created this
model_version.
- current_stagestring
Current stage of the model version. Valid values are
None,Staging,ProductionandArchived. A new model version starts inNone.
- descriptionstring
User-provided description of the model version.
- sourcestring
URI of the model artifacts that the model version was registered from, as given in the create request.
- run_idstring
ID of the MLflow run that produced the model artifacts, if
sourcecomes from a run in the MLflow tracking server.
- statusstring
Registration status of the model version. The version can be used once the status is
READY.PENDING_REGISTRATIONFAILED_REGISTRATIONREADY
- status_messagestring
Details on the current
status, such as the reason why registration failed.
- tagsarray of object
Key-value tags attached to the model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
URL of the MLflow run that generated the model version, as given in the create request. It is typically set when the run is hosted on a different tracking server than the registry.
Create a model version GA
POST
1 more path
POST
Creates a model version.
API scopes: mlflow
Request body
- namestringRequired
Name of the existing registered model to create the version under.
- sourcestringRequired
URI of the model artifacts to register, such as the artifact path of an MLflow run or a
models:/<model_name>/<version>URI of an existingREADYmodel version. The registry copies the artifacts asynchronously, so the new version starts inPENDING_REGISTRATIONstatus.
- run_idstring
ID of the MLflow run that produced the model artifacts, if
sourcecomes from a run in the MLflow tracking server. The ID is at most 32 characters long.
- tagsarray of object
Tags to set on the new model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
MLflow run link - this is the exact link of the run that generated this model version, potentially hosted at another instance of MLflow.
- descriptionstring
Description of the model version.
Response
- model_versionobject
The new model version, including the version number that the registry assigned.
Show child attributesHide child attributes
- namestring
Name of the registered model that the version belongs to.
- versionstring
Version number of the model version, as a string. Version numbers start at 1 for each registered model.
- creation_timestampint64
Time when the model version was created, as a Unix timestamp in milliseconds.
- last_updated_timestampint64
Time when the metadata of the model version was last updated, as a Unix timestamp in milliseconds.
- user_idstring
User that created this
model_version.
- current_stagestring
Current stage of the model version. Valid values are
None,Staging,ProductionandArchived. A new model version starts inNone.
- descriptionstring
User-provided description of the model version.
- sourcestring
URI of the model artifacts that the model version was registered from, as given in the create request.
- run_idstring
ID of the MLflow run that produced the model artifacts, if
sourcecomes from a run in the MLflow tracking server.
- statusstring
Registration status of the model version. The version can be used once the status is
READY.PENDING_REGISTRATIONFAILED_REGISTRATIONREADY
- status_messagestring
Details on the current
status, such as the reason why registration failed.
- tagsarray of object
Key-value tags attached to the model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
URL of the MLflow run that generated the model version, as given in the create request. It is typically set when the run is hosted on a different tracking server than the registry.
Update model version GA
PATCH
1 more path
PATCH
Updates the description of a model version.
API scopes: mlflow
Request body
- namestringRequired
Name of the registered model
- versionstringRequired
Model version number
- descriptionstring
New description for the model version.
Response
- model_versionobject
The updated model version.
Show child attributesHide child attributes
- namestring
Name of the registered model that the version belongs to.
- versionstring
Version number of the model version, as a string. Version numbers start at 1 for each registered model.
- creation_timestampint64
Time when the model version was created, as a Unix timestamp in milliseconds.
- last_updated_timestampint64
Time when the metadata of the model version was last updated, as a Unix timestamp in milliseconds.
- user_idstring
User that created this
model_version.
- current_stagestring
Current stage of the model version. Valid values are
None,Staging,ProductionandArchived. A new model version starts inNone.
- descriptionstring
User-provided description of the model version.
- sourcestring
URI of the model artifacts that the model version was registered from, as given in the create request.
- run_idstring
ID of the MLflow run that produced the model artifacts, if
sourcecomes from a run in the MLflow tracking server.
- statusstring
Registration status of the model version. The version can be used once the status is
READY.PENDING_REGISTRATIONFAILED_REGISTRATIONREADY
- status_messagestring
Details on the current
status, such as the reason why registration failed.
- tagsarray of object
Key-value tags attached to the model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
URL of the MLflow run that generated the model version, as given in the create request. It is typically set when the run is hosted on a different tracking server than the registry.
Delete a model version GA
Search model versions GA
GET
Searches for specific model versions based on the supplied filter.
API scopes: mlflow
Parameters
- filterstringquery
Filter condition on
name,run_id,sourceortags.<key>. The filter must constrain at least one ofname,run_idorsource. Conditions use=,!=,LIKEorILIKE, andrun_idalso supportsIN. String values are wrapped in single quotes, and several conditions can be combined withAND.
- max_resultsint64<= 10000query
Maximum number of models desired. Max threshold is 10K.
- order_byarray of stringquery
Columns to order the results by, each given as
name,version,current_stageorlast_updated_timestampwith an optionalASCorDESCsuffix, whereASCis the default. Tiebreaks are done by latest stage transition timestamp, followed by name ASC, followed by version DESC.
- page_tokenstringquery
Pagination token to go to next page based on previous search query.
Response
Returns a list of ModelVersion objects.
Get a model version URI GA
GET
Gets a URI to download the model version.
API scopes: mlflow
Parameters
- namestringRequiredquery
Name of the registered model
- versionstringRequiredquery
Model version number
Response
- artifact_uristring
URI corresponding to where artifacts for this model version are stored.
Set a version tag GA
POST
1 more path
POST
Sets a model version tag.
API scopes: mlflow
Request body
- namestringRequired
Unique name of the model.
- versionstringRequired
Model version number.
- keystringRequired
Name of the tag. Maximum size depends on storage backend. If a tag with this name already exists, its preexisting value will be replaced by the specified
value. All storage backends are guaranteed to support key values up to 250 bytes in size.
- valuestringRequired
String value of the tag being logged. Maximum size depends on storage backend. All storage backends are guaranteed to support key values up to 5000 bytes in size.
Delete a model version tag GA
DELETE
Deletes a model version tag.
API scopes: mlflow
Parameters
- namestringRequiredquery
Name of the registered model that the tag was logged under.
- versionstringRequiredquery
Model version number that the tag was logged under.
- keystringRequiredquery
Name of the tag. The name must be an exact match; wild-card deletion is not supported. Maximum size is 250 bytes.
Transition a stage GA
POST
1 more path
POST
Transition a model version's stage. This is a <Databricks> workspace version of the MLflow endpoint that also accepts a comment associated with the transition to be recorded.
API scopes: mlflow
Request body
- namestring[ 1 .. 100 ] charactersRequired
Name of the model.
- versionstringRequired
Version of the model.
- stagestringRequired
Target stage of the transition. Valid values are:
-
None: The initial stage of a model version. -
Staging: Staging or pre-production stage. -
Production: Production stage. -
Archived: Archived stage.
-
- archive_existing_versionsbooleanRequired
Specifies whether to archive all current model versions in the target stage.
- commentstring[ 1 .. 65535 ] characters
User-provided comment on the action.
Response
- model_version_databricksobject
Updated model version
Show child attributesHide child attributes
- namestring[ 1 .. 100 ] characters
Name of the model.
- versionstring
Version of the model.
- creation_timestampint64
Creation time of the object, as a Unix timestamp in milliseconds.
- last_updated_timestampint64
Time of the object at last update, as a Unix timestamp in milliseconds.
- user_idstringemail
The username of the user that created the object.
- current_stagestring
Current stage of the model version. Valid values are
None,Staging,ProductionandArchived.
- descriptionstring<= 65535 characters
User-specified description for the object.
- sourcestring
URI that indicates the location of the source model artifacts. This is used when creating the model version.
- run_idstringuuid
Unique identifier for the MLflow tracking run associated with the source model artifacts.
- statusstring
Registration status of the model version. The version can be used once the status is
READY.PENDING_REGISTRATIONFAILED_REGISTRATIONREADY
- status_messagestring
Details on the current status, for example why registration failed.
- open_requestsarray of object
Open stage transition requests for the model version.
Show child attributesHide child attributes
- creation_timestampint64
Creation time of the object, as a Unix timestamp in milliseconds.
- user_idstringemail
The username of the user that created the object.
- activity_typestring
Type of activity. Valid values are:
-
APPLIED_TRANSITION: User applied the corresponding stage transition. -
REQUESTED_TRANSITION: User requested the corresponding stage transition. -
CANCELLED_REQUEST: User cancelled an existing transition request. -
APPROVED_REQUEST: User approved the corresponding stage transition. -
REJECTED_REQUEST: User rejected the coressponding stage transition. -
SYSTEM_TRANSITION: For events performed as a side effect, such as archiving existing model versions in a stage.
APPLIED_TRANSITIONREQUESTED_TRANSITIONCANCELLED_REQUESTAPPROVED_REQUESTREJECTED_REQUESTNEW_COMMENTSYSTEM_TRANSITION
-
- commentstring[ 1 .. 65535 ] characters
User-provided comment associated with the activity, comment, or transition request.
- last_updated_timestampint64
Time of the object at last update, as a Unix timestamp in milliseconds.
- from_stagestring
Source stage of the transition (if the activity is stage transition related). Valid values are:
-
None: The initial stage of a model version. -
Staging: Staging or pre-production stage. -
Production: Production stage. -
Archived: Archived stage.
-
- to_stagestring
Target stage of the transition (if the activity is stage transition related). Valid values are:
-
None: The initial stage of a model version. -
Staging: Staging or pre-production stage. -
Production: Production stage. -
Archived: Archived stage.
-
- system_commentstring
Comment made by system, for example explaining an activity of type
SYSTEM_TRANSITION. It usually describes a side effect, such as a version being archived as part of another version's stage transition, and may not be returned for some activity types.
- available_actionsarray of string
Array of actions on the activity allowed for the current viewer.
APPROVE_TRANSITION_REQUESTREJECT_TRANSITION_REQUESTCANCEL_TRANSITION_REQUESTEDIT_COMMENTDELETE_COMMENT
- idstringuuid
Unique identifier for the object.
- permission_levelstring
Permission level of the requesting user on the registered model that the version belongs to.
CAN_MANAGECAN_EDITCAN_READCAN_MANAGE_STAGING_VERSIONSCAN_MANAGE_PRODUCTION_VERSIONSCAN_CREATE_REGISTERED_MODEL
- tagsarray of object
Array of tags that are associated with the model version.
Show child attributesHide child attributes
- keystring
The tag key.
- valuestring
The tag value.
- run_linkstring
URL of the run associated with the model artifacts. This field is set at model version creation time only for model versions whose source run is from a tracking server that is different from the registry server.
- email_subscription_statusstring
Email Subscription Status: This is the subscription status of the user to the model version Users get subscribed by interacting with the model version.
ALL_EVENTSDEFAULTSUBSCRIBEDUNSUBSCRIBED
- feature_listobject
Feature lineage of
model_version.Show child attributesHide child attributes
- featuresarray of object
Show child attributesHide child attributes
- feature_table_namestring
Feature table name
- feature_namestring
Feature name
- feature_table_idstring
Feature table id