Band teaser with your own track: which Sume models take audio

To score a teaser with your own track, use a model whose catalog lists audio references: wan-3.0, minimax-h3, minimax-h3-max or a Seedance id. gemini-omni-flash-1.1 has reference_audio_urls disabled and Kling 3 is not listed among the models that accept reference inputs. A 12-second clip costs $0.90 on H3 at 768p, $1.50 on Wan 3.0 at 720p, $4.54 on Seedance 2.0 and $6.94 on Seedance 2.5.

Send the track as an audio_url in input_references, with a trimmed excerpt, not the full song. Source: Video generation and Video Router, read 2026-10-06.

What does a 12-second teaser cost?

Twelve seconds, 720p (H3 at 768p).

How do I attach the track?

An audio_url entry sits beside any image references. Only models that list audio references accept it; others answer with a 400.

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": "Band on a dark stage, handheld camera, lights pulsing with the beat",
    "duration": 12,
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "input_references": [
      {"type": "audio_url", "audio_url": {"url": "https://example.com/hook-excerpt.mp3"}}
    ]
  }'

Does the track have to be short?

Trim it to the section you want under the clip. A 12-second teaser needs 12 seconds of audio, not the whole song.