Kling 4.0 multiple keyframes: ten, versus Sume's two frames

Kling 4.0 announces up to 10 flexible keyframes on one timeline. Sume's documented frame inputs are two: first_frame and last_frame in frame_images (or image_url plus end_image_url on the Omni route). There is no documented ten-keyframe input on Sume, so for a longer chain you generate clips in sequence and join them.

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

What are Kling's Multiple Keyframes?

The post says Kling 4.0 "supports up to 10 flexible keyframes, allowing you to define important visual states at different points on the timeline", and recommends using them to guide actions, camera changes and story turns.

What frame inputs does Sume document?

The model listing's supported_frame_images field says which frame_type values a model accepts, and the docs' example shows ["first_frame", "last_frame"]. On the Omni route the image-to-video row takes image_url plus an optional end_image_url.

How do I chain more than two frames?

Generate one clip per pair of anchors. Use a last frame of clip one as the first_frame of clip two; Video frames extracts a still at a time you name, unbilled, and Video trim returns a new MP4 holding only [start, end). Join the results with Timeline 1.0, as in assembling long-form video.

Will the join look continuous?

Not guaranteed. Sume's docs do not promise continuity across separately generated clips, so check the seam. Review the joined sequence and regenerate a segment rather than assuming a ten-keyframe path. For a model with a different keyframe limit, see Luma Ray 3.2 keyframes.