3-minute Instagram Reel: total cost by build route on Sume

A 3-minute Instagram Reel built from generated clips costs about $23.65 on the cheapest video routes here (18 Gemini Omni Flash clips or six Wan 3.0 clips, both at 720p, which tie) and about $105.16 on the most expensive (six 30-second Seedance 2.5 clips at 720p). The spread is almost entirely the clip model, because narration, captions, render and music add the same $1.15 to every route.

Instagram's 3-minute ceiling is the reason to price this at all: Metricool's Instagram news page reports Instagram's Reels guide listing clip lengths from 3 seconds to 3 minutes (read 2026-10-03). Every figure below is Sume's own billable rate, which is the provider list price times 1.25. The arithmetic is shown so you can swap in your own clip count.

Assumptions you can change

A 180-second Reel is built as one narrated spine. The script is assumed to run 2,700 characters, which is 15 characters per second; that ratio is our assumption, not a Sume figure, so replace it with your measured script length. Clips are 9:16.

Clip lengths follow the Video Router limits in the docs: Seedance 2.5 takes 4 to 30 seconds per clip, Wan 3.0 takes 2 to 30, and Gemini Omni Flash 1.1 takes 3 to 10. So a 180-second cut needs at least six Seedance or Wan clips, or at least 18 Omni clips.

• Narration: Sume TTS bills 1 credit per character; at the pricing package's list credit price that is $0.000038 per character, or $0.0475 per 1,000 after the 1.25 margin.
• Captions: a flat $0.20 per accepted job for videos up to 60 seconds, so three jobs cover three 60-second pieces.
• Timeline 1.0: $0.10 per ceil(output minute), so 180 seconds is 3 minutes and $0.30.
• Music: $0.125 per accepted generation on the Music Router.

Finishing costs the same on every route

Finishing is the part people forget. It is small next to the clip bill, but it is also the part that does not shrink when you pick a cheaper model. This table shows the unrounded narration and total; the route table below rounds to cents.

Total by build route

The generation column is pieces times the per-piece price. Seedance 2.5 at 720p bills $0.5778 per second, so a 30-second clip is $17.334; Wan 3.0 at 720p is $0.1250 per second, so 30 seconds is $3.75; Omni at 720p is $0.1250 per second, so 10 seconds is $1.25.

The still-image route is not a video model at all: 36 Nano Banana 2 stills at $0.10 each, held 5 seconds apiece in the Timeline. It is the cheapest build, and it is also a slideshow, so only use it when the narration carries the Reel.

Price the retakes before you start

Generated video is rarely one take per slot. The table scales only the generation column, because a retake does not repeat the narration, caption or music job. At 1.5x, one failed clip in every two slots is retaken once.

The takeaway is that retakes cost far more on the expensive route: an extra half of the Seedance 2.5 720p bill is $52.00, while an extra half of the Omni bill is $11.25.

What Sume does and does not do here

Sume prices each step as its own job and reserves the amount on submit. It does not generate one 3-minute video in a single call, since no listed video model takes more than 30 seconds, so the Timeline job is what joins the pieces. Billable rates are read from the catalog; confirm live numbers in GET /v1/catalog before quoting a client.

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 = [
    ('Wan 3.0 720p clips, 180 s', 180, 0.125),
    ('Narration characters', 2700, 0.0000475),
    ('Caption jobs', 3, 0.20),
    ('Timeline minutes', 3, 0.10),
    ('Music Router track', 1, 0.125),
]
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}")