WanVideoFree

About WanVideoFree

WanVideoFree is a small, independently run tool for generating short videos with hosted Wan 2.2, 2.5 and 2.7 endpoints plus a few comparison models. It is not Alibaba's official Wan website.

That last part shapes everything on this site, so it's worth being explicit about it.

What we actually are

A thin, honest layer on top of hosted inference. We don't train models. We don't own GPUs. What we do:

  • Give you a browser interface to video models without requiring local weights or a compatible GPU
  • Pick sensible defaults so you don't have to learn every parameter
  • Tell you what things cost, before you spend anything

Why we talk about cost so much

Most "free AI video" sites don't mention money at all. We do, on nearly every page, because the number explains everything else:

ModelReal hosted cost
Wan 2.5 · 480p~$0.05 / second
Wan 2.5 · 1080p~$0.15 / second
Wan 2.2 · 720p T2V~$0.08 / video on the selected 5B endpoint
Wan 2.7 · 720p~$0.10 / second

A five-second 1080p clip therefore costs about $0.75 in GPU time before anyone earns anything.

This is why:

  • Starter credits are limited rather than unlimited
  • The lowest-cost draft uses a specific Wan 2.2 endpoint
  • We'd rather show you the ceiling than hide it behind a paywall you discover later

If a site promises unlimited free HD video, check how long it has been online.

What we do when something goes wrong

Generation fails sometimes — a prompt gets blocked by the model's content filter, a provider times out, a queue backs up. When that happens:

  • Your credits are returned. Automatically, not on request.
  • You get told why, in plain language, not a generic "generation failed."

We learned this the hard way: an early user was charged credits for a generation that never ran, and saw no explanation. That's been fixed, and the fix is the reason this section exists.

What we don't do

  • We don't animate photos of real people without their consent, and we block obvious abuse
  • We don't add hidden charges — credit cost is shown before you generate
  • We don't claim quality we can't deliver; each page says what the model is bad at, not just what it's good at

Models we run

All current generation runs on fal.ai as the inference provider. The site's topical focus is Wan; other catalog entries exist for real model choice and comparisons. Official Wan2.1 and Wan2.2 repositories publish open model resources, while hosted Wan 2.5 and 2.7 access is described as hosted capability rather than automatically labeled open source.

Contact

Questions, bug reports, refund issues: huangxiaofeng775@gmail.com