Runs overview - Chainshift

A run is an asynchronous collection job that executes your registered prompts on AI models. When you submit one, a runID is returned immediately and collection proceeds in the background. Check for completion by polling (every 30 seconds is recommended).

Submission rules

Item	Rule
Prompt selection	Exactly one of `promptIDs` (1 to 500) or `promptSetID`
`models`	Values from List supported models only. At least 1; duplicates count once
`region`	Optional. Defaults to `KR`
`location`	Optional. A code from List supported locations. When omitted, the run uses the `region` level. ChatGPT and Perplexity only
`repeatCount`	1 to 100. Repeats the same combination (for measuring visibility rate)
Total runs	prompts × models × repeats ≤ 10,000 (`RUN_TOO_MANY_TASKS`)

Statuses

Run request: only RUNNING → FINISHED. Even if individual runs fail, the run request ends as FINISHED; check the failure count in failedCount.
Individual run: PENDING → SUCCESS or FAILED.

Billing

Credits are charged after completion, not at submission: each run that finishes collection is charged at the per-model rate. Failed runs are not charged. See Credits.

FAQ

Where do I get the answer body? The answers field in the run status only contains progress. For the body and sources, query List answers with the runID filter. Why does the run list return only 10 items at a time? The limit for GET /runs is at most 10 (other lists allow 100). Iterate with the cursor if you need to retrieve many runs. See the Running prompts guide for the full flow and polling code.