Seven persona avatars, one 10-second clip each: $19.53 at standard
Seven reusable avatars at $0.95 each plus a 10-second video apiece: $19.53 at standard, $23.80 at plus on Sume. Handles and workflow.
Seven avatars, each with one 10-second talking video, cost $19.53 at standard (19.530) and $23.80 at plus on Sume (as of 2026-10-09). Each avatar is $0.95 to create. Each video is 10 seconds x $0.184 at standard ($1.84) or x $0.245 at plus ($2.45).
The table
The scenario is a set of seven personas for a sales deck or a training course, each introducing themselves once. Creation and rendering are separate jobs.
| Item | Arithmetic | Standard | Plus |
|---|---|---|---|
| 7 avatars | 7 x $0.95 | $6.65 | $6.65 |
| 7 videos of 10 s | 7 x 10 x rate | $12.88 | $17.150 |
| Total | sum | $19.53 | $23.800 |
Three ways to define a persona
Avatar creation takes one of three inputs: a prompt (describe the person), props (structured traits: ethnicity, sex and age) or a photo (a public HTTPS image URL). All three cost the same $0.95. If your seven personas come from a spreadsheet of traits, props is the easiest to automate; prompt gives more freedom; photo is the one that needs rights to the face in the picture, which is a decision to make before the call, not after.
Each avatar gets a handle you choose, such as product_host. Sume normalizes a leading @ and stores the handle without it. The handle goes into the talking-video request as avatar_handle.
Reuse is where it pays
The $0.95 is paid once per avatar, so reuse lowers the per-video cost. The same seven avatars with three 10-second videos each at standard cost 6.65 + 21 x 1.84 = $45.29, which is $2.157 per video. With one video each it is $2.79 per video.
Plan the persona list around reuse: a persona that will appear in a series is worth creating; one used once could be a prompt-only avatar with a short handle. Treat the handle as an asset name so it is easy to find in the avatar list later.
Run order and checks
Submit the seven creation jobs with different Idempotency-Key values and poll each one. Only start a video when its avatar job has completed. If one creation fails, you do not pay for it and the other six are unaffected. The avatar docs describe the job lifecycle and the resource routes (list and read) that let you confirm which handles are ready.
A tip for the spreadsheet: put the handle, the cost and the job id in the same row.
Checking the plus figure: 6.65 + 7 x 2.45 = $23.80, the same as 7 x (0.95 + 2.45) = 7 x 3.40. Verify totals by the row arithmetic in the table before you quote them.
- Use one Idempotency-Key per avatar, such as avatar-persona-1 to avatar-persona-7.
- Keep a note of each handle next to the source of its face.
- Render the first video at standard to check the face before spending on plus or max.
Sources
Related posts
More in Sume Avatar 1.0
- Three 15 s avatar hooks at standard ($8.28) vs one at max ($8.25)
On Sume Avatar 1.0, three 15 s hooks at standard cost 3 x 15 x $0.184 = $8.28, about the same as one 15 s render at max (15 x $0.55 = $8.25).
- AI doctor or expert on screen in a TikTok Shop health ad: banned
TikTok Shop's AI content page prohibits AI-fabricated expert or doctor identities endorsing health products. What to film, and what an avatar may still do.
- What to log for a likeness consent trail: five fields per request
Sume avatar and face swap requests carry an avatar_handle, a source URL, a quality, an Idempotency-Key and a job id. Log those five with the consent record.
- Which MCP grant lets an agent create an avatar? Write, no paid scope
On Sume's hosted MCP, avatars_create and the avatar video tools are paid tools that need the Write toggle at consent. There is no mcp:paid scope.
Written by Sume