Out-of-office video message with an AI avatar: 15 seconds on Sume

A 15-second out-of-office video from an avatar costs $2.76 on Standard, $3.68 on Plus and $8.25 on Max, with no product image and before the fixed $0.20 caption add-on (shown in the table). Write the message as three beats (who is away, until when, who to contact), submit one talking-video job, and link the finished MP4 from your auto-reply.

The prices come from Sume's public catalog and the avatar video guide, read 2026-10-04. The point of a video here is to be clearer than a text auto-reply, not to replace it, so keep the text version too.

A 15-second script that fits

Sume estimates duration from the script and accepts 4 to 60 seconds, so 15 seconds is comfortably inside the window. A short script has room for only three facts; leave out the apology and the story.

• Beat 1 (about 3 seconds): who you are and that you are away.
• Beat 2 (about 5 seconds): the date you are back, spoken as a weekday and a date.
• Beat 3 (about 7 seconds): the person or address to contact instead, said slowly enough to write down.

Cost by tier

A one-off message is a good fit for Standard: nobody studies an auto-reply for polish.

Turn captions on

Most people open a message without sound at work. Add the captions object so the words are burned into the MP4; the style slam is the default for Latin text.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: ooo-2026-10-12" \
  -d '{"avatar_handle": "team_host", "quality": "standard", "script": "Hi, this is the product team. I am away until Monday the twelfth. For anything urgent, write to support at example dot com.", "captions": {"enabled": true, "style": "slam", "language": "auto"}}'

What can go wrong

A caption failure does not fail the video: the job can succeed with a clean video_url and captions.status=failed, and you can re-caption it as a standalone job. A return date that changes after you render means a new job, because the spoken words are baked in.

Say in the clip, in the caption or in the message around it that the presenter is AI-generated. The API does not add that label for you, so put it in the script or the page the video sits on.