Submit annotations for an annotation queue record.
Annotations are upserted into the underlying data source by annotation config name. Omitted annotation configs are left unchanged.
Payload Requirements
annotations must contain at least one entry.name must match an annotation config associated with the queue.label, score, or text to leave the existing value unchanged. Individual fields cannot be set to null; annotations cannot be removed once written.Response Returns a snapshot of the fields updated by this operation: the record identity and the submitted annotations only. Evaluations and user assignments are not included for performance reasons. Use the list records endpoint to retrieve the full record state.
Valid example
{
"annotations": [
{"name": "accuracy", "label": "correct", "score": 1.0},
{"name": "quality", "text": "Well-structured response"}
]
}
Invalid example (annotation name not in queue)
{
"annotations": [
{"name": "unknown_config", "label": "good"}
]
}
Most Arize AI endpoints require authentication. For those endpoints that require authentication, include your API key in the request header using the format
The unique identifier of the annotation queue A universally unique identifier
"RW50aXR5OjEyMzQ1"
The unique identifier of the annotation queue record A universally unique identifier
"RW50aXR5OjEyMzQ1"
Body containing annotations to submit for an annotation queue record
Annotations to submit for an annotation queue record. Annotations are upserted by annotation config name; omitted configs are left unchanged.
Annotations to upsert on this record, keyed by annotation config name. There is no maximum limit — you may submit one annotation per annotation config associated with the queue.
1Returns a snapshot of the record fields updated by the annotate operation. Only the submitted annotations are included. Evaluations and user assignments are omitted; use the list records endpoint for the full record state.
A snapshot of the annotation queue record fields that were modified by an annotate operation. Only the record identity fields and the submitted annotations are returned. Evaluations and user assignments are not fetched and are not included in this response for performance reasons; use the list records endpoint to retrieve the full record state.
The unique identifier for the record
The annotation queue this record belongs to
The source type of the record (spans or dataset)
spans, dataset The annotations that were submitted in this request