Dog photo to a 5-second clip with sound: six Sume models
Animate one dog photo for 5 seconds at 720p. Six Sume video ids priced side by side, from $0.38 to $2.89, and which of them cannot turn the sound off.

Sending one dog photo as the first frame for a 5-second clip costs $0.38 on minimax-h3 (768p native), $0.63 on wan-3.0 or gemini-omni-flash-1.1 at 720p, $0.94 on seedance-2-mini, $1.05 on kling-3 with sound and $2.89 on seedance-2.5. Those are Sume's billable prices: the catalog rate times 1.25, rounded up to the cent.
Two of these always return sound. Omni Flash 1.1 and H3 have native audio with no toggle, and generate_audio: false is a 400 on Omni. If you want to lay your own music over the clip, pick a model that accepts the toggle, or trim the audio afterwards. Source: Video Router, read 2026-10-06.
What does each cost for the same photo?
Same input, same 5 seconds, each model at the closest 720p rung it offers.
| Model id | Resolution | Length | Sume price | Note |
|---|---|---|---|---|
| minimax-h3 | 768p | 5 s | $0.38 | sound always on |
| wan-3.0 | 720p | 5 s | $0.63 | audio supported |
| gemini-omni-flash-1.1 | 720p | 5 s | $0.63 | sound always on; 16:9 or 9:16 |
| seedance-2-mini | 720p | 5 s | $0.95 | |
| kling-3 | 720p | 5 s | $1.05 | with sound; $0.70 silent |
| seedance-2.5 | 720p | 5 s | $2.89 |
How do I send it?
Put the photo in frame_images as the first_frame. A single photo in input_references instead is reference-to-video, which conditions the whole clip on the dog but does not pin the opening frame. That split is how /v1/videos avoids the old single-image ambiguity, described on Video generation.
curl -sS -X POST https://api.sume.com/v1/videos \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax-h3",
"prompt": "The dog turns its head and wags its tail, soft window light",
"duration": 5,
"resolution": "768p",
"frame_images": [
{"type": "image_url", "image_url": {"url": "https://example.com/dog.jpg"}, "frame_type": "first_frame"}
]
}'What should I watch for?
Run the cheapest row first. A pet photo with a clear face and one simple motion rarely needs the $2.89 tier; move up only if the cheap take loses the dog's markings. Sume reserves the price when you submit, so check the balance before queueing a batch.
Sources
Related posts
More in Use cases
- Dub a 12-minute video into 4 languages: STT once, TTS four times
Dubbing a 12-minute video into four languages costs about $1.84 on Sume for transcription and speech, if each script is 9,000 characters. Translation is extra.
- Dub a short into Spanish with an API: detach, transcribe, speak, swap
A Spanish dub on Sume is six jobs you control: detach the audio, transcribe with sentence segments, translate yourself, speak each line, join, and render.
- Product hero video: pin the first frame to your own packshot
Use frame_images with first_frame on sume/auto so a 6-second product video opens on your exact packshot, not a regenerated lookalike.
- EU Article 50 AI avatar video disclosure checklist for marketers
Tavus says EU Article 50 has required disclosure of AI likenesses since 2 August 2026. A checklist of what to label on an avatar clip, with the Sume fields.
Written by Sume