Caption 25 shorts in 3 languages: 75 jobs at $0.20 is $15

Captioning 25 shorts that each exist in three languages is 75 caption jobs, because the speech differs per language file. Sume's standalone video captions endpoint charges $0.20 per accepted job for a video up to 60 seconds under the current fixed estimate, so the total is 75 × $0.20 = $15.00. A restyle from source_caption_id is billed as a render too, so changing the look later costs the same $0.20 again, though it skips a second transcription.

The arithmetic

Prices are from the Sume docs and the OpenAPI summary for POST /v1/video-captions; confirm the live number in GET /v1/catalog before a big batch.

Why the restyle is not free

The captions page says that passing source_caption_id instead of video_url reuses the earlier caption's source video and word timings, so no second speech-to-text runs. It also says billing is unchanged, because a restyle is still a render. So pick the style on a pilot before the batch: one short in three styles costs 60 cents and saves you $15 of re-burns.

Order of work

Put captions last. They burn into pixels, so any later trim, reorder or voice change means burning again. The usual order is voiceover, then cut, then captions; the voiceover for shorts post walks through it.

For a batch, submit with mode: "webhook" so you are told when each job reaches a terminal state, and keep polling as a backup, as the webhooks page recommends. Use a distinct Idempotency-Key per clip and language so a retry does not buy a caption twice.

Compared with a streaming transcript

Microsoft's MAI-Transcribe-2-Streaming costs $0.54 per hour of audio on an introductory rate through the end of 2026 (read 2026-10-04). 25 shorts of 45 seconds is about 19 minutes of audio per language, so that is a few cents of recognition. But recognition is only the transcript. The video captions job also renders the styled words onto your video, which is what the $0.20 buys.