sdk-v2

Create Eval Run

Enqueue a persisted benchmark run and execute it in the background.

The run row (snapshots + context) is the durable artifact: the response returns immediately and the report is polled via GET /eval-runs/{id}.

post/v2/sdk/eval-runs

Request

  • Base URL: https://api.pandas-ai.com (the document declares no server; this is the origin it was published from)
  • URL: https://api.pandas-ai.com/v2/sdk/eval-runs
  • Auth: none declared

Request body

suite_idstring required

Benchmark to run

versionstring nullable

Semantic-layer version; default Current

tagsstring[] nullable

Selection tags (disambiguate entries)

languagestring nullable

Response language code, e.g. 'it'

disambiguationboolean nullable

Expose structured ask questions (mirrors the SDK flag)

rowsinteger

Shadow rows per table

saltstring

Reroll the synthetic world

behaviorinteger nullable

Agent behavior major to measure; default the newest served

retention_modestring

What the persisted report keeps: bounded evidence or metadata only

case_idsstring[] nullable

Subset of case ids; default all

Response

Successful Response

unknown required

Changes

No recorded changes to this operation across all 1 revision of this API.