Square pottery clips at 12 s: Wan 3.0 $1.50 vs Seedance 2.5 $6.93

A 12-second 1:1 clip costs $1.50 on Wan 3.0 at 720p and $6.93 on Seedance 2.5 at 720p on Sume, a 4.6x gap. Both accept 12 seconds and a square frame, so the choice is price against look (read 2026-10-09).

Same request, two models

Both models list 1:1 and any whole number of seconds in their range (Wan 3.0 2 to 30, Seedance 2.5 4 to 30). That lets you A/B the same prompt by changing only model.

A/B in one script

Submit both jobs with different idempotency keys, then compare the two outputs and usage.cost on the completed responses.

import os, requests

H = {"Authorization": "Bearer " + os.environ["SUME_API_KEY"]}
prompt = "A potter shapes a bowl on a wheel, warm studio light"
for model in ('wan-3.0', 'seedance-2.5'):
    body = {'model': model, 'prompt': prompt, 'duration': 12,
            'resolution': '720p', 'aspect_ratio': '1:1'}
    r = requests.post('https://api.sume.com/v1/videos',
                      headers={**H, 'Idempotency-Key': f'ab-{model}-001'},
                      json=body, timeout=60)
    print(model, r.status_code, r.json().get('polling_url'))

A 40-clip catalog

For 40 square product clips at 12 seconds and 720p, Wan 3.0 costs 40 x $1.50 = $60.00, and Seedance 2.5 costs $277.34. At 480p the two are $30.00 and $123.26.

The gap is large enough that a two-stage plan makes sense: draft on Wan 3.0 at 480p to check composition, then decide whether Seedance 2.5 is worth the extra for the hero clips.

Seedance 2.5 and Wan 3.0 share the 1:1 ratio, whole-second durations and the same POST /v1/videos request, so switching between them is a one-field change.

What to check

• Seedance 2.5 reaches 1080p at $1.42 per second for square; Wan 3.0 stays at $0.25 per second.
• Judge on your own prompt. A price ratio does not tell you which clip looks better.
• Both models accept image, video and audio references, so the same reference set works in both runs.
• The cheapest square test is Wan 3.0 at 480p: $0.75 for 12 seconds.