Hedra timeline editing vs Sume timeline: 10 cents a minute

Sume's timeline API lets you assemble clips, audio and captions in a JSON body: POST /v1/timeline-1.0/plan is unbilled and POST /v1/timeline-1.0/render costs $0.10 per output minute, rounded up. Hedra's page lists timeline editing among its updates but I found no price for it, so I can only compare the shape of the workflow, not the cost.

What the two offer

Hedra's description is a single line on a page dated June 1, 2026. The Sume details come from the Timeline docs.

Plan before you render

The plan call is the useful habit. It returns the estimate and any refusal before you spend, so run it for every edit.

• Build the timeline body with your clips, an optional voice-over and captions.
• Call /v1/timeline-1.0/plan and read the estimate and any warnings.
• Trim anything that pushes you past a minute boundary: 61 seconds bills two minutes, and 181 bills four.
• Call /v1/timeline-1.0/render with an Idempotency-Key so a retried request cannot render twice.
• Fetch the result from GET /v1/jobs/:id/result.

A worked example

Take a 3-minute product explainer with four clips, a voice-over and captions. The plan call is free, so you can run it five times while you adjust the cut. The render is one output of 180 seconds, which bills three minutes: $0.30. If a last-minute change pushes the cut to 181 seconds, the same render bills four minutes, $0.40, so a trim of one second saves ten cents. That is small for one video and visible across a hundred. The stored posts on billing minutes show the same rule with a 61-second short that bills two minutes.

The audio side is separate. If your voice-over is several takes, POST /v1/timeline-1.0/audio concatenates up to 20 parts into one track, and you pass that track to the render. Keep each take as its own file until the last moment, so replacing one line does not mean regenerating the rest.

What Sume does not do

Sume's timeline is not a visual editor. You do not drag clips on a canvas; you describe the edit as data. That suits automated pipelines, where the same body is built by code, and it suits review, because the body is a file you can diff. If your team wants to scrub a preview by hand, a UI-based editor such as the one Hedra lists is the closer fit, and you can still use Sume to produce the clips that go into it.

Mixed frame rates are another point to watch. When clips from different models are joined, set the output frame rate on purpose rather than leaving it to a resample, as the stored posts on judder explain.