Skip to main content
POST
Legacy Dataset, Evaluation, and Report endpoints are deprecated for new workflows. Use the Tables API for new dataset import, evaluation, scoring, recalculation, and reporting workflows.
Create a dataset version asynchronously from existing request logs. You can populate the version with explicit request log IDs or with structured request-log filters.

​
Behavior Notes

  • If both request_log_ids and filter_group are present, explicit request IDs take precedence.
  • The endpoint queues a background job and reuses the draft dataset for the dataset group when one already exists.
  • Jobs are capped at 50,000 request logs.
  • Filter-based datasets persist their filter parameters so refresh flows can replay the same query later.
  • Completion triggers the dataset_version_created_from_filter_params webhook.

Authorizations

​
X-API-KEY
string
header
required

Body

application/json

Dataset version creation payload from request IDs or request-log filters.

​
dataset_group_id
integer
required

Dataset group that will receive the generated dataset version.

​
request_log_ids
integer[] | null

Explicit request log IDs to snapshot. When provided, this mode takes precedence over filter_group.

Maximum array length: 50000
​
filter_group
StructuredFilterGroup · object | null

Structured request-log filters used to populate the dataset asynchronously.

​
q
string | null

Free-text request-log search query applied with filter_group.

​
sort_by
enum<string> | null
Available options:
request_start_time,
input_tokens,
output_tokens,
cost,
latency_ms,
status
​
sort_order
enum<string> | null
Available options:
asc,
desc
​
variables_to_parse
string[] | null

Request fields or variables to parse into dataset columns.

​
limit
integer | null

Maximum number of request logs to include. When set, only the first N matching rows (in the current sort order) are added to the dataset version.

Required range: x >= 1

Response

Dataset version creation queued.

​
success
enum<boolean>
required
Available options:
true
​
dataset_id
integer
required
​
dataset_group_id
integer
required
​
version_number
integer
required

Version number of the draft or generated dataset version.

​
message
string