Kling 4.0 video editing: five video inputs, Sume edits one

Kling 4.0 editing is announced with up to five video inputs, one of them primary, combined up to 30 seconds. Sume's edit route, gemini-omni-flash-1.1 with video_to_video, takes a single video_url, and the docs say that video_url cannot be combined with image_url, end_image_url or reference_*_urls.

Kling facts are from its blog post (full model "coming in October 2026"); Sume facts are from Video Router and the API catalog, read 2026-09-30.

What does Kling 4.0 editing accept?

The post says the full model "supports up to five video inputs in an editing task, including one primary video, with a combined duration of up to 30 seconds". It describes adding, replacing or removing elements, changing backgrounds and adjusting details while preserving the rest of the scene.

What does a Sume edit request take?

The Omni edit row is video_to_video with video_url; the prompt describes the edit, resolution is optional (default 720p), and aspect_ratio and duration are not accepted. The catalog also notes that duration, if given, is only a reserve-estimate hint.

Can I add reference images to an edit on Sume?

Not on the same request. The docs state video_url is the edit source, not a reference. If you want reference-driven generation from clips, use reference_video_urls on a model that accepts them; Omni takes up to 3, each 3 seconds or shorter. Another catalog model, a motion-transfer one, takes exactly one source video plus 1 to 8 images on its own route.

How do I run a multi-clip edit today?

Edit each clip in its own request and join the results with Timeline 1.0. For prompt wording see Gemini Omni video edit prompts, and note the regional limit covered in uploaded video edits.