Short notes on what the individual knobs of a generative video request cost, and on the limits that read like preferences but behave like budgets.
These notes come from the team behind wan-3.run, an independent third-party browser interface to Alibaba's Wan 3.0 video model. We are not affiliated with Alibaba or Alibaba Cloud, and that is stated on every page here and on the product itself.
We have a commercial interest, which is exactly why every factual claim here carries a source and a check date. You should not have to trust a vendor-adjacent site; you should be able to verify it. Where we could not verify something against Alibaba's own documentation, it is not on these pages.
Because the pricing shape of this model is per second of output, and that inverts some habits carried over from per-request APIs. On a per-request endpoint you batch to save money. Here, batching saves round trips but not money, resolution is a multiplier rather than a setting, and the cheapest useful thing you can do — iterate at 480P — is not the default.
None of that is hidden. It is all in the published parameters. It just is not what most material about this model chooses to write about.
A large share of what currently ranks for this model is wrong in specific, checkable ways: a 4K tier that does not exist, a 21:9 ratio that is not in the list, 24 fps when the output is 30, and open weights that were never announced. The Wan 3.0 specification table exists so there is something to point at instead — every published parameter, each one sourced and dated, including the unflattering ones.
If you find an error on any page here, it is worth telling us. A specification page that is wrong is worse than no page at all.