Azure Foundry sora-2 2025-12-08 retires Oct 15: an 11-day plan
The last Azure Foundry Sora deployment, sora-2 version 2025-12-08, retires 2026-10-15 with no replacement listed. Eleven days from today: what to do each day.

Microsoft's retirement schedule lists sora-2 version 2025-12-08 as Preview with a retirement date of 2026-10-15 and no replacement model. From today, 2026-10-04, that is 11 days. Plan to point your video calls at a model that is still being sold, and test it before the date, because the schedule offers no successor to migrate to.
What the schedule says (read 2026-10-04)
The table copies the two sora-2 rows on the Foundry page, whose own last-updated date is 2026-09-21. Read the page again before you act; schedules change.
| Model | Version | Lifecycle | Retirement date | Replacement listed |
|---|---|---|---|---|
| sora-2 | 2025-10-06 | Preview | 2026-07-15 | sora-2 (2025-12-08) |
| sora-2 | 2025-12-08 | Preview | 2026-10-15 | None |
OpenAI's own API is already gone
OpenAI's deprecations page says the Videos API and the sora-2 and sora-2-pro ids were removed on 2026-09-24, with no recommended replacement. So the Azure deployment is the last place the model name still resolves, and it has 11 days left.
An 11-day plan
Spend the first days on inventory and the last days on a cutover you can reverse.
- Days 1 to 2: list every deployment, job and script that names a Sora deployment, and note the clip length, aspect ratio and whether you rely on audio.
- Days 3 to 4: map each use to a model in
GET /v1/videos/modelsby duration, resolution and audio, not by brand. - Days 5 to 7: render the same ten prompts on the candidate and compare them side by side.
- Days 8 to 9: switch a small share of traffic behind a flag, keeping the old path for rollback.
- Days 10 to 11: switch fully and remove the Sora code path.
Pointing the same prompt at Sume
Sume's POST /v1/videos uses the OpenRouter video wire, so the request body is a small change. Model ids are bare, for example seedance-2 or wan-3.0. Sume does not carry Sora, so this is a replacement, not a drop-in: expect a different look and re-test your prompts.
curl -X POST "https://api.sume.com/v1/videos" \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: sora-move-001" \
-d '{
"model": "seedance-2",
"prompt": "A paper boat crossing a rain puddle, shallow focus",
"duration": 8,
"resolution": "720p"
}'What Sume does not do
Sume has no Sora model, no Azure deployment type and no Foundry quota. It has a bare-id catalog, an async job envelope with polling_url, and webhook or polling completion. If a compliance rule needs Azure residency, that is a separate decision.
Sources
Related posts
More in Models
- Azure sora-2: July 15, then October 15, and no successor listed
Azure listed sora-2 2025-10-06 to retire 2026-07-15 and sora-2 2025-12-08 to retire 2026-10-15. Only the first has a replacement. How to read a version chain.
- ChatGPT picks Flare or Sunburst for you; the API does not
In ChatGPT, Flare handles standard use and Sunburst takes complex prompts. In the API you choose. How to build that rule on Sume, where sume/auto uses Flare.
- Choose an AI video model by the asset you already have: Sume ids
Photo, two frames, product shots, a voice track, a driving clip or a finished video: which Sume video id fits each starting asset, from the catalog rows.
- Swap the gift in a Christmas clip with Gemini Omni Flash edit
Gemini Omni Flash 1.1 on Sume edits a 3 to 10 second clip by prompt, so one approved Christmas ad can become three gift variants without a reshoot.
Written by Sume