Veo 3.1 clip older than 2 days: it cannot be extended
Veo 3.1 extends only Veo-made clips kept 2 days. If yours expired, here is what Google's page allows and what Sume's Omni edit and new generation do.

If a Veo 3.1 clip is older than 2 days, you cannot extend it through the Gemini API: Google's Veo page says generated videos are stored for 2 days at most, and that only Veo-generated videos are eligible for extension. The page describes no route for extending a file you re-upload. Your options are to regenerate the opening, or to continue the story with a new clip in another model. Facts below were read on 2026-10-08.
Sume does not list Veo 3.1, so it has no Veo extension. What it lists for the same job is Gemini Omni Flash 1.1, which generates 3 to 10 seconds per request and has a separate video_url edit mode.
What Google's Veo page says about extension
The numbers come from the Veo 3.1 page. They are the reason the 2-day window matters.
| Rule | Value |
|---|---|
| Step length | 7 seconds per extension |
| Extensions allowed | Up to 20 times |
| Input clip | At most 141 seconds, 720p only |
| Output length | Original plus extension, up to 148 seconds |
| Eligible clips | Only Veo-generated videos |
| Storage | 2 days; an extension resets the timer |
Three ways to continue an expired clip
First, regenerate. Write the prompt of the lost clip again and extend within the window this time. The cost is the original generation again; Google's pricing page lists Veo 3.1 Fast at $0.10 per second for 720p, so an 8-second redo is $0.80, and Standard is $0.40 per second, $3.20 for 8 seconds.
Second, treat the last frame as a new first frame. Export the final frame of the downloaded clip and use it as the first frame of a new clip. On Sume, gemini-omni-flash-1.1 accepts image_url for that and returns 3 to 10 seconds with native audio, billed at provider list times 1.25 per second: $0.125 at 720p, so $1.00 for 8 seconds.
Third, edit instead of extend. Sume's Omni row has a video_url mode: the prompt is the instruction, resolution is optional and defaults to 720p, and you do not send aspect_ratio or duration for it. That changes what is in the clip, but it does not add running time.
Cost comparison for an 8-second continuation
Using the same 8 seconds at 720p, the numbers are close enough that the deciding factor is the join, not the price.
| Route | Per second | 8 seconds |
|---|---|---|
| Veo 3.1 Fast (Google direct) | $0.10 | $0.80 |
| Veo 3.1 Standard (Google direct) | $0.40 | $3.20 |
| Gemini Omni Flash 1.1 on Sume | $0.125 | $1.00 |
Keep the next one safe
Download every Veo clip the day you make it, and keep the prompt with it. On Sume, a finished job gives you a media.sume.com artifact URL; save the file if it is a deliverable. If you need the join to look continuous, use the last frame as the first frame and describe the same camera move in the prompt. Both models add native audio, so check the seam by ear.
A habit that avoids the problem
Download every Veo file the day you make it. Google's Veo page says videos are stored for 2 days and that extending resets the timer, so a clip you plan to extend should be extended inside that window. For a clip that has expired, regenerate it, since the page describes no other route.
Veo 3.1 Lite is $0.05 per second at 720p, which makes a 8-second redo $0.40.
Sources
Related posts
More in Models
- Video editing board: Wan 3.0 1,187, Seedance 2.5 1,151; Sume uses Omni
Wan 3.0 tops the AA video-editing board at 1,187 Elo. Sume's documented edit route is video_url on gemini-omni-flash-1.1, third at 1,139. Price inside.
- AI video with native audio: which Sume models let you turn it off
On Sume, only Kling v3 Pro prices audio separately (+$0.07/s). Gemini Omni and MiniMax always make audio; Grok Imagine and Genjutsu make none. Table inside.
- Vidu Q4 takes 15 image refs: what Sume's reference limits are
Vidu Q4 Preview lists up to 15 image and 3 audio references. Sume does not list Vidu; Gemini Omni Flash 1.1 takes 10 images and 3 clips of 3 s, none audio.
- Wan 2.2 CUDA out of memory: offload_model, dtype and t5_cpu flags
The Wan 2.2 README fixes OOM with three flags: offload_model, convert_model_dtype and t5_cpu. Which commands need 80 GB and 24 GB, and what it does not promise.
Written by Sume