Online course intro bumper: 6 seconds at 1080p on Sume

A 6-second 1080p course-intro bumper costs $0.84 on kling-3 silent, $1.13 on gemini-omni-flash-1.1, $1.20 on minimax-h3-max, $1.26 on kling-3 with sound, $1.50 on wan-3.0 and $8.53 on seedance-2.5 on Sume. The bumper is a good fit for pinning a first frame: design the title card as an image, send it as first_frame, and let the model animate the background around it.

That second part matters because video models can garble small text. Put the course title in your own image, not in the prompt. Rates and limits are from Video generation and Video Router, read 2026-10-06.

What does the bumper cost per model?

Six seconds at 1080p, 16:9.

How do I keep the title legible?

Send the finished title card as the first frame and write a prompt that moves only the background: slow parallax, drifting light, no camera push into the text. Keep the last frame free: if the title must hold to the end, send the same image as last_frame on a model that lists end-frame support.

curl -sS -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "wan-3.0",
    "prompt": "Soft light drifts behind the title card, no camera movement, text stays sharp",
    "duration": 6,
    "resolution": "1080p",
    "frame_images": [
      {"type": "image_url", "image_url": {"url": "https://example.com/title-card.png"}, "frame_type": "first_frame"},
      {"type": "image_url", "image_url": {"url": "https://example.com/title-card.png"}, "frame_type": "last_frame"}
    ]
  }'

Does the same bumper work for every lesson?

One bumper is one job. Generate it once and reuse the file; the Sume timeline can join it to lesson clips. Do not re-render per lesson unless the title card changes, because every render is a new charge.