AI video launches, week of Oct 4, 2026: what Sume lists
One table of this week's video news: Kling 4.0 Flash, Tavus Griffin-Lite, Runway Seedance 2.5 Draft, Veo 3.1 shutdown. Which each maps to on Sume, with caveats.
Short answer
Four things moved in the last week and only some of them reach an API buyer. Kling 4.0 Flash is a closed beta, Tavus Griffin-Lite is a research preview for select testers, Runway added Seedance 2.5 Draft mode, and Google's Veo 3.1 preview ids have a shutdown date of Oct 22. On Sume you can call seedance-2.5 and gemini-omni-flash-1.1 today; you cannot call Kling 4.0, Griffin or Veo.
The week in one table
| Item | Status | On Sume |
|---|---|---|
| Kling 4.0 Flash | Closed beta since Sep 28, per fal | Not listed; kling-3 is |
| Tavus Griffin-Lite | Research preview, trusted testers, announced Oct 1 on Tavus | Avatar video jobs, not a live-conversation model |
| Runway Seedance 2.5 Draft | Added Oct 2 per the Runway changelog | seedance-2.5 accepts 480p |
| Veo 3.1 preview ids | Shutdown Oct 22 per Google | Not listed; gemini-omni-flash-1.1 is |
How to act on each
- Kling: do not build on a model with no API price; use
kling-3and keep the model id in config. - Griffin: nothing to integrate yet. Tavus says it is not available to customers at this time.
- Runway Draft: if you want cheap 480p Seedance 2.5 drafts without a Runway plan, send
resolution: "480p"toseedance-2.5. - Veo: if you hold a veo-3.1 preview id, Google points to gemini-omni-1.1-flash; on Sume the equivalent id is
gemini-omni-flash-1.1.
A safe pattern for any launch week
Do not hard-code a launch. Read the catalog, then pick a pinned id or sume/auto as covered in pin an id or use auto:
curl https://api.sume.com/v1/videos/models \
-H "Authorization: Bearer $SUME_API_KEY"What none of this changes
Launch weeks reward calm. A model that is in a closed beta has no stable id, no stable price and no stable limits, so any code you write against it is a guess. A model that is listed in the Sume catalog has a published id, supported resolutions, durations and reference types, and a quote you can read before you spend.
If your goal is a 30 second clip, the catalog already has options: seedance-2.5 accepts 4 to 30 seconds and wan-3.0 accepts 2 to 30 seconds. If your goal is a short clip with synced audio at 1080p, gemini-omni-flash-1.1 accepts 3 to 10 seconds. None of those is Kling 4.0, but they are available now, and each can be swapped later by changing one string.
Caveats
Status above is as of 2026-10-04 and changes quickly. Kling 4.0 details come from fal's guide because Kling's release-notes page did not render its body. Sume's catalog is the source of truth for what you can call.
Sources
Related posts
More in Models
- 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.
- 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.
Written by Sume