Test the Wan 3.0 leaderboard yourself: 6 prompts, 4 models, $12.84

Wan 3.0 and Seedance 2.5 are reported to top the video leaderboard (reported by Pinggy, read 2026-10-07). A leaderboard averages other people's prompts. Six of your own, rendered by four models, tells you more and costs $12.84. Prices are Sume billable rates (provider list x 1.25, rounded up to the cent per job), read 2026-10-07.

Cost of one prompt across four models

Judge blind: have a colleague rate the 24 clips without model names. Keep the same prompt, duration and aspect ratio (16:9 is accepted by all four). Seedance 2.5 is a fifth candidate priced per 1000 tokens, so it is not in this table.

The runnable estimate:

from decimal import Decimal as D, ROUND_CEILING
rates = {"wan 720p": "0.125", "h3 768p": "0.075",
         "h3max 768p": "0.10", "omni 720p": "0.125"}
def job(r, s):
    return (D(r)*s*100).to_integral_value(ROUND_CEILING)/100
one = sum(job(r, 5) for r in rates.values())
print(one, one*6)