OpenRouter batch results kept 30 days; Sume media URLs don't expire

OpenRouter deletes batch inputs and results 30 days after creation, then returns 410 Gone. Sume job results use durable media.sume.com URLs that do not expire.

4 min readSume
All posts

OpenRouter's Batch API keeps batch inputs and results for 30 days after creation, after which retrieving a completed batch returns 410 Gone. Sume's Format runs docs say media URLs are durable media.sume.com HTTPS URLs that do not expire, so there is no stated download deadline to schedule around. The docs state this for Format runs; the jobs pages say to use the Sume media URLs from a result.

Sources: OpenRouter's Batch API quickstart and Sume's Format runs and Jobs and results pages, read 2026-10-01.

How long does OpenRouter keep batch results?

The quickstart says OpenRouter stores batch inputs and results as JSONL artifacts and deletes them 30 days after creation, and that upstream retention varies by provider. It advises downloading anything you need before the window ends, or deleting the batch sooner to purge it. Billing, generation and audit records are retained.

How long do Sume results last?

The Format runs page states that media URLs are durable media.sume.com HTTPS URLs and that they do not expire. The same page warns that they are public to anyone holding the URL, so proxy or copy them if you need per-customer access control. The jobs docs add (without a separate retention statement) that you should use the Sume media URLs from the result, not raw provider URLs, which are not public API outputs.

Retention, read 2026-10-01: https://docs.sume.com/formats/runs
QuestionOpenRouter Batch APISume job results
Retention clock30 days after batch creationNo clock stated; Format run URLs do not expire
After the windowCompleted batch returns 410 GoneNot applicable
AccessResults returned inline in the batch responsePublic media.sume.com URL
Early removalDELETE on a terminal batchNot described in these pages

Do I still need to store anything?

Yes: store the job id. GET /v1/jobs/:id/result answers only for completed jobs, and the API reference lists GET /v1/jobs to list workspace jobs filterable by status and type, so a lost id can be recovered. Because the media URL is public, copy files you must protect. For the same question about video, see do Sume video URLs expire.

What should a migration change?

If a pipeline copies OpenRouter results out before day 30, keep that step for OpenRouter and drop the deadline for Sume jobs. Persist the result URL next to the job id once the job is completed.

Sources

Related posts

More in Developers

All Developers posts

Written by Sume