Together AI says download videos immediately: Sume media URLs
Together's docs say not to rely on video URLs for long-term storage. Sume returns media.sume.com artifacts as its public contract. What to save in each case.
Together AI tells you to download a finished video immediately and not to rely on its URL for long-term storage; Sume documents media.sume.com artifact URLs as the public contract for results. The Together instruction is explicit, but the page fetched on 2026-10-02 gives no retention number, so treat the URL as temporary and copy the file.
Together's guidance is on its video generation overview; Sume's is on Jobs and results, Core workflow and the API reference.
What exactly does Together say?
After a job reaches completed, the video URL is read from outputs.video_url. The page's note is short: download videos immediately after completion and do not rely on the URLs for long-term storage. It does not state how long the URL stays valid, so any number you see elsewhere is not confirmed by this page.
The safe pattern is to treat the URL as a hand-off: fetch the bytes into your own bucket in the same step that sees completed.
What does Sume promise about result URLs?
Sume's Core workflow page says completed jobs can include public artifacts under https://media.sume.com, that Sume-owned artifact URLs are the public contract, and that raw provider URLs are not. The API reference repeats that public results should use media.sume.com URLs and that raw provider URLs are not exposed.
Those pages do not state a retention period for artifacts, so Sume's docs do not let you claim one either. If a deliverable has a legal or contractual shelf life, keep your own copy there too.
How do the two compare?
The difference is what the docs commit to, not a measured lifetime.
| Question | Together AI | Sume |
|---|---|---|
| Where is the URL? | outputs.video_url on a completed job | result_url, then the artifacts in the result |
| Host | Not stated on the page | media.sume.com |
| Stated retention | None on the page | None in these docs |
| Instruction | Download immediately | Use Sume artifact URLs, not provider URLs |
What should a pipeline do in both cases?
Do the download inside the completion handler. On Sume that handler is either your poller, which should stop on terminal and read result_url once result_ready is true, or a signed job.completed webhook with polling kept as a backup. Store the job id with the file so a re-download is a lookup, not a resubmit; resubmitting a paid job bills again.
Sume also lets you reuse media by URL in later requests, since launch requests accept public HTTPS media URLs. That is a convenience for chaining, not an archive.
Sources
Related posts
More in Comparisons
- Together AI video status values vs Sume job statuses (canceled)
Together's video job has five statuses, including cancelled. Sume has five too, but spells it canceled and adds terminal and result_ready flags. Map them here.
- Veo 3.1 1080p 8-second clip cost vs Sume per-second rows
What an 8-second 1080p clip costs on Veo 3.1 Standard, Fast and Lite list prices versus five 1080p rows in the Sume video catalog, dated 2026-10-01.
- Veo 3.1 4K price per second vs Gemini Omni Flash 4K
Veo 3.1 4K costs $0.60 per second on Standard and $0.30 on Fast. Gemini Omni Flash 1.1 on Sume lists 4K at $0.375 per second. Dated 2026-10-01.
- Veo in Google Ads: 3 images, 10 seconds, and the Sume equivalent
Veo in Google Ads takes up to three images and returns videos up to 10 s. Here is the same job on Sume's video router, with its limits, for use outside Google.
Written by Sume