1-hour transcript cost: $0.60 on Sume, two 30-minute files

A one-hour transcript costs $0.60 on Sume at the public rate of $0.01 per audio minute: 60 minutes times one cent. Because a clip passed to Video inspect must be 1,800 seconds or shorter, an hour is two uploads of up to 30 minutes each, and the two pieces add to the same 60 billable minutes.

That one-cent rate is the provider list price of $0.008 per audio minute for ElevenLabs Scribe v2 times the 1.25 house margin. The table below scales it from ten minutes to a year of weekly shows. Figures were read from the pricing package and the Video inspect docs on 2026-10-03.

Where the $0.01 comes from

In the pricing package the fixed SKU for an inspect transcript is described as billed per ceil(audio minute), with one minute reserved when no duration hint is sent. The same page in the docs states the public rate as $0.01 per audio minute and tells you to confirm it live in GET /v1/catalog.

Two caps shape how you plan an hour. The source clip may not exceed 1,800 seconds, and the optional duration_seconds hint tops out at 600. The hint sizes the reservation and the docs say capture never exceeds the reservation, so treat the table as a budget at the documented rate and compare it with the settled amount on a first run.

Splitting an hour: the rounding example

Minutes round up per file, not across the whole recording, so how you cut matters at the margin. A 59 minute 30 second recording cut into 1,800 and 1,770 second files bills 30 plus 30 minutes. If you cut it into six 10-minute pieces plus fractions, each piece rounds up on its own.

Cutting is a separate step. Video trim is $0.02 per job, but it also refuses a source longer than 1,800 seconds, so for an hour-long master you cut the two halves in your own editor before you upload; trim is for shortening a file that is already under the cap.

What else is on the bill

The transcript line is not the whole inspect bill. The docs say every inspect also reserves its Modal compute ceiling and is captured at container seconds times the Modal list rate times 1.25, plus a platform fee, never above the hold. The transcript adds its per-minute rate to that reservation. So $0.60 is the transcript charge at the documented rate; compute for the job is billed on top.

If you only want text and no stills, send frames: false, which the inspect docs describe for a probe-only call.

What Sume does and does not do

Sume returns text and words[] and, with segmentation.mode: "sentence", gapless sentence segments. It does not label speakers, and it will not take an hour in one file. Use the table to budget, then watch the Usage page for the settled amount.

• Rate: $0.01 per audio minute, rounded up per file.
• Cap: source 1,800 s; duration hint 600 s.
• Language: send language_code as a hint or let it auto-detect.

Rerun the arithmetic

The same line items as a short script, so you can change a quantity and see the total. It is plain arithmetic on the rates in this post; it does not call the Sume API.

# Sume billable USD rates, read 2026-10-03. Edit the quantities.
lines = [
    ('Transcript, audio minutes', 60, 0.01),
]
total = 0.0
for label, qty, unit in lines:
    cost = qty * unit
    total += cost
    print(f"{label:<46}{qty:>8g} x {unit:<10g} = ${cost:,.2f}")
print(f"{'Total':<46}{'':>8}   {'':<10}   ${total:,.2f}")