Cost to render a 30-minute video on Sume: a $3.00 ceiling

Rendering a 30-minute video on Sume's Timeline costs at most $3.00. The public rate is $0.10 per output minute, rounded up, the render reserves that on submit, and 1,800 seconds is the longest output a single render accepts. The charge is captured at the render's own compute and never exceeds the reservation.

That ceiling is for assembly only: Timeline joins clips you already have on Sume with an audio track, with no provider inference. The cost of generating the clips is separate and is priced per model.

What does each length cost?

The reserve is ceil(audio.duration_seconds / 60) minutes at $0.10, per the Timeline docs. Because the audio length defines the output length, you know the reserve from one number before you submit.

Rounding is up, not to the nearest minute, so a 61-second render reserves two minutes. If you can trim a final second off to land on a minute boundary, the render is cheaper.

What if my video is longer than 30 minutes?

audio.duration_seconds accepts 1 to 1,800. Longer than that is outside one render. Split the program at a natural break, render two parts, and join them with whatever tool you use downstream. Two 30-minute parts reserve at most $6.00 together.

Timeline also takes 1 to 200 video slots, and every media URL must already be a media.sume.com artifact or asset in your workspace, so import external files first with a media import.

How do I get the number without spending?

Use the plan call. POST /v1/timeline-1.0/plan runs the schema and host checks and the compiler and returns billable_minutes and estimated_cost_usd_micros. The docs say it creates no job, reserves nothing, downloads no media and does not need an idempotency key. It cannot predict padding or loop warnings for short sources.

A plan response with billable_minutes: 12 should line up with the $1.20 row above. The request below uses the audio and video fields from the render example; if the plan call rejects a field, its error names it.

curl -X POST https://api.sume.com/v1/timeline-1.0/plan \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"audio":{"url":"https://media.sume.com/artifacts/artf_demo/voice.wav","duration_seconds":720},"video":[{"source_url":"https://media.sume.com/artifacts/artf_demo/talk.mp4","start":0,"duration":720}]}'

What happens to the reservation after the render?

Rendering is billed in two steps. Submit reserves the ceiling. Success captures the render's own compute at list times 1.25 plus a platform fee, never above the reservation. A failed render releases the whole reservation. Errors and credits lists 402 insufficient_credits for a balance that cannot cover the reserve.

So the practical rule for budgeting is simple: assume the reserve, expect a little less. A 30-minute program needs $3.00 of balance available at submit, not $3.00 spent.

How does a Timeline render compare with generating the clips?

Assembly is the cheap part of a long video. A 30-minute program costs at most $3.00 to join, while generating even a few minutes of clips with a video model costs more per minute of footage than the whole assembly costs per minute.

That is useful for budgeting. If you reuse the same clips in several cuts, you pay for generation once and for assembly each time, at $0.10 per output minute. A re-cut with a different audio track is a new render, not a new generation.

The same logic applies to short formats. A 24-second render is $0.10, so trying three versions of an ad costs thirty cents of assembly on top of the clips.