Eight avatars on Sume: $7.60 to create, $117.60 for 24 plus clips
Eight Sume Avatar 1.0 characters cost 8 x $0.95 = $7.60 to create; three 20-second plus videos each (24 x 20 s x $0.245) add $117.60, $125.20 in all.
A cast of eight Sume Avatar 1.0 characters costs 8 x $0.95 = $7.60 to create. If each character then appears in three 20-second videos at the plus tier, that is 24 videos x 20 seconds x $0.245 = $117.60, so the whole cast and its first round of videos costs $125.20.
Creation is a one-time line item
The catalog prices avatar creation at $0.95 per avatar, described as one-time creation for reusable Avatar 1.0 video workflows. The docs offer three input types: a text prompt, structured profile traits (the props type) and a reference image (the photo type). The catalog lists a single creation price, so the three paths cost the same.
Each request needs a top-level avatar_handle. A leading @ is accepted and stored without it, so pick handles that are stable keys in your own system.
Plan the handles before you create anything. A leading @ is stripped on storage, so @product_host and product_host are the same handle. Eight distinct handles also keep your records tidy: one row per handle with the creation job id, the input type, and the date.
| Line | Quantity | Unit price | Cost |
|---|---|---|---|
| Avatar creation | 8 | $0.95 | $7.60 |
| Plus video seconds | 480 (24 x 20 s) | $0.245 | $117.60 |
| Total | $125.20 |
Video seconds dominate
Creation is 6 percent of the total here ($7.60 of $125.20). The lever is the per-second tier: the same 480 seconds at standard cost 480 x $0.184 = $88.32, and at max 480 x $0.55 = $264.00. Choose creation count by how many distinct faces you need on screen, and choose the tier by how the clips will be used.
The preview flow helps at cast scale. A first-frame preview is a still; it shows the composition before the video render. For 24 videos at 20 seconds, a single wasted plus render costs 20 x $0.245 = $4.90, so checking previews pays for itself after the first caught mistake.
- One resolved avatar appears in each final video, so a two-person scene needs two jobs.
- Use the preview flow to approve first frames before spending on a render.
- Send an Idempotency-Key on every create and render so a retry does not duplicate a charge.
Photo avatars and permission
When you create an avatar from a reference image, the image_url must be a fetchable public HTTPS image. The docs rule on URL safety, not on whose face is in the photo, so keeping a record of who agreed to the use of each likeness is your process, not something the API checks.
Naming and ownership deserve a line in your process. Each avatar is a stored resource you can list with GET /v1/avatar-1.0/avatars and read by id, so a weekly job can reconcile the list against your own register of who is who. If an avatar is no longer wanted, stop using its handle in new requests; the docs describe creation and reading, and they do not describe a price for keeping an avatar, so the cast has no running cost beyond the renders.
Sources
Related posts
More in Sume Avatar 1.0
- Face swap beta, 9-second clip: you reserve the 15-second price
Sume's Avatar Face Swap beta reserves the maximum 15-second source price whatever your clip length: $2.76 at standard, $3.675 at plus, $8.25 at max.
- Griffin-Lite's 0.27 to 0.59 s latency in 25 fps frames vs Sume clips
Tavus reports Griffin-Lite video latency of 0.43 s average (0.27 best, 0.59 worst). At 25 fps that is 7 to 15 frames. Sume renders recorded clips.
- Narrate a listing in your avatar's voice: TTS with avatar_handle
Sume TTS 1.0 can speak with the voice of a ready avatar if you pass avatar_handle. A 1,200-character listing script costs $0.057.
- One 15-second avatar script in 9:16, 1:1 and 16:9: cost of all three
Rendering one 15 s script in three aspect ratios is three jobs: $8.28 at standard, $11.025 at plus, $24.75 at max. Aspect ratio does not change the price.
Written by Sume