6-second face swap clip: $1.10 standard, $1.47 plus, $3.30 max

A 6-second source clip through Sume's Avatar Face Swap beta costs $1.104 at standard, $1.470 at plus and $3.30 at max (as of 2026-10-09). The hold shown on your balance rounds the standard figure up to $1.11. The rates are $0.184, $0.245 and $0.55 per second of source video.

The three quality tiers

In the beta, quality is a required field. The endpoint does not pick a default for you: if you omit it, the request is refused. The docs list the same three values that Avatar Video uses: standard, plus and max.

Constraints a 6-second clip passes

The face-swap docs describe the working window as approximately 4 to 15 seconds with usable audio, so 6 seconds is inside it. The source must be a public HTTPS video URL. The endpoint rejects localhost, private-network, non-HTTPS, signed or private URLs, and provider task URLs. It accepts no prompt, transcript, duration knob or aspect ratio: the request is the avatar handle, the video URL and the quality.

You also need a ready avatar. Creating one costs $0.95 once, so a first swap is 0.95 + 1.104 = $2.054 at standard.

• Body: avatar_handle, video_url, quality (all required).
• Route: POST /v1/models/sume/avatar-face-swap/v1.0/runs.
• Result: a media.sume.com video URL once resource_status is ready.

Job handling

A face swap job is asynchronous. The endpoint supports async, sync or subscribe with wait_timeout_seconds, and webhook with a public HTTPS webhook_url. For a 6-second clip you can ask the endpoint to wait, then fall back to polling the status URL. When the resource is ready, resource_status is the primary signal, and the result carries a media.sume.com video URL.

A clip that fails validation, for example one with no usable audio or outside the roughly 4 to 15 second range, fails with a public error and the reservation is refunded. Keep the source URL public and stable until the job finishes, because a signed or private URL is rejected up front.

Try standard first

Because the clip is short, run standard first and look at the result. Upgrading to max for the same clip costs 6 x (0.55 - 0.184) = $2.196 more. If the face holds up under standard, the other two tiers are not worth buying for a 6-second reaction shot.

A last number to keep: five 6-second swaps at standard are 5 x 1.104 = $5.52, so a small batch fits inside a single $10 top-up with the avatar.

Finally, decide what you will do with the audio. The docs describe the endpoint as taking no transcript or prompt, so the spoken words stay those of the source clip. If you need new words from the avatar, that is a talking-video job, not a swap, and 6 seconds of it at standard is 6 x 0.184 = $1.104, the same price by coincidence of rate.