What a 30-second Short costs to finish: trim, captions, render

Finishing a 30-second vertical clip with Sume costs $0.32 at the public rates in the docs: $0.02 for a trim, $0.20 for captions and $0.10 for a timeline render. The probe and the timeline plan are free, and the cost of generating the clip is separate.

The itemized bill

Each number is the public rate on its own docs page, and each page says to confirm the live price on GET /v1/catalog. The sum is plain addition; there is no discount for chaining the steps and no extra platform fee that I read in these pages.

What is not in the $0.32

What the total does not include. The generation step is billed per model at the provider list rate times 1.25, so its price depends on the model, resolution and duration you choose. Read the live rate from the catalog for your model. A transcript from video inspect is optional and costs $0.01 per audio minute, with a one minute reserve when you do not send duration_seconds.

Skip what you do not need

Not every step is needed. A clip that already has the right length needs no trim. A clip that came with the right audio and no need for captions skips that $0.20. A single clip with no cuts may not need a timeline at all. The itemized bill is a ceiling for the full chain, not a minimum.

Where rounding applies

Rounding matters in one place only. Timeline bills per output minute, rounded up, so a 30 second render and a 59 second render both cost $0.10, and a 61 second one costs $0.20. Captions have a fixed estimate for videos of at most 60 seconds, so that line does not move with a 30 second clip.

Order the calls

Use the free steps to protect the paid ones. This sequence checks the file first, so you only pay for the paid calls on a clip you have already looked at:

# 1. free: probe the clip
curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: finish-probe-001" \
  -d '{"video_url":"https://media.sume.com/artifacts/artf_demo/raw.mp4","frames":false}'

# 2. free: plan the render, read billable_minutes
# 3. $0.02: trim   4. $0.20: captions   5. $0.10: render

Fail before you pay

A failed request is not the same as a charge. The captions docs say a bad style or an out-of-range value gives a 400 at request time, and that you do not pay for an incorrect render. Keep the Idempotency-Key fixed on a retry so a repeat call returns the same job.

Scope of the number

This is the cost of the finishing steps in the Sume docs, not a claim about the cost of a campaign. Ad spend and creative testing on a platform are outside it.

Scale it to a batch

Scale the arithmetic to a batch. Ten finished clips at the same chain cost ten times $0.32, or $3.20, before generation. That multiplication is the only math in this post, and it works because each step has a flat rate: the trim and the caption are per job, and the render is per rounded-up minute. If your clips differ in length, recompute the render line only.

A second caption style costs a second job

Compare the chain with doing a step twice. Restyling captions with source_caption_id reuses the source video and its word timings, so speech-to-text does not run again, but the docs say the price does not change because a restyle is still a render. So a second look costs another $0.20; pick the style before you burn, using the free stills and the style list.

Record the rates you read

Keep one line in your records for each finished clip: the three job ids and the three rates you read that day. Rates can change, and the pages themselves say to confirm on the catalog. A dated record makes your own cost numbers checkable later without relying on this post.

Why it is worth checking

A cost line in a plan is only useful if it matches the bill. After your first real batch, compare your sum with what the account shows for the same jobs. If they differ, the live catalog or the job result will tell you which rate moved, and you can fix the estimate before the next batch.

The takeaway

At the public rates, the full finishing chain for a 30-second clip is $0.32, and every number should be checked against the live catalog.