Skip to content
The differentiator

Open-source video generation, transparently priced

Wan 2.2, LTX-2.5, HunyuanVideo and more — the video model pricing landscape other providers won’t show you.

From $0.017 per second of generated video — that’s roughly $0.09 for a five-second clip.

The price opacity problem

Video generation pricing is one of the least transparent corners of the AI API market. The same underlying model can cost 10x more depending on where you access it — and plenty of providers won’t quote a rate at all until you have an account. We publish every rate, side by side, so you can see exactly what you’re paying for.

Wan 2.2 T2V A14B
$0.029/s
Wan 2.2 I2V A14B
$0.032/s
LTX-2.5
$0.024/s
HunyuanVideo 1.5
$0.027/s

Filled bar: our rate. Full bar: the highest published rate we found for the same model elsewhere.

What we don’t resell, and why

We host and resell inference for open-source video models under their respective open licenses only. We do not resell access to closed-weight video models. Third-party resale of closed-source model access frequently violates the origin provider’s terms of service — a risk we don’t take on, and don’t want our customers exposed to either.

  • Sora 2(OpenAI)Closed weights — no resale license exists
  • Veo 3.1(Google DeepMind)Closed weights — no resale license exists
  • Runway Gen-4(Runway)Closed weights — no resale license exists
  • Kling 2.5(Kuaishou)Closed weights — no resale license exists
  • Wan 2.5 / 2.6(Alibaba)API-only — weights never released, despite the open Wan 2.1/2.2 lineage

Worth being precise about the last one: Alibaba open-sourced Wan 2.1 and 2.2, and we serve those. Wan 2.5 and 2.6 are API-only — the weights were never released — so anyone reselling them is not reselling an open model, whatever the lineage of the name suggests.

Every video model, every rate

Priced per second of generated output. No per-job minimum, no resolution surcharge.

Text models priced per 1M tokens · video per second of output · image per generated image.

Submit a job, poll for the result

The cost comes back with the job id, before any GPU time is spent.

import requests

job = requests.post(
    "https://router.xark.io/api/v1/video/generations",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "model": "alibaba/wan-2.2-t2v-a14b",
        "prompt": "a paper boat drifting down a rain-slicked street at night",
        "duration": 5,
        "resolution": "720p"
    }
).json()

print(job["id"], job["estimated_cost_usd"])   # ← cost known before it renders

See it for yourself

Get an API key and run a model in about twenty seconds.

Video rendering requires an API key. Text models run without one.