CogVideoX-5B API
6s max · 480p / 720p · Released Aug 27, 2024 · By Zhipu AI · Apache 2.0
CogVideoX-5B costs $0.017 per second, against an official rate of $0.03 — 43% below official pricing. It is a Zhipu AI model under the Apache 2.0 licence, callable through an OpenAI-compatible endpoint with the model id `zhipuai/cogvideox-5b`. This model is listed for reference and is not currently served by an upstream provider, so requests for it return no available channel.
Not currently available. No configured upstream serves this model, so a request for it returns no available channel. The rate below is what it will cost when it is enabled.
Input
$0.017/sec
$0.030/sec
Billing
per second
no per-job minimum
Savings
43%
off official
- No markup
- OpenAI SDK compatible
- No minimum spend
- Apache 2.0
The cheapest video generation on the platform. Older and lower resolution than the rest, but hard to beat when you need thousands of short clips.
Copy this and go
The model id is already filled in. Nothing in this snippet needs editing except your key.
import requests
job = requests.post(
"https://router.xark.io/api/v1/video/generations",
headers={"Authorization": "Bearer YOUR_API_KEY"},
json={
"model": "zhipuai/cogvideox-5b",
"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
CogVideoX-5B pricing across providers
Rates as published by each provider. We update these when they change; if you spot one that's stale, tell us and we'll fix it.
| Provider | Input | Output |
|---|---|---|
| AI Token Router | $0.017/sec | — |
| Official (Zhipu AI) | $0.030/sec | — |
| Replicate | $0.038/sec | — |
| fal.ai | $0.035/sec | — |
| OpenRouter | Not offered | — |
What it's good at
- Bulk clip generation on a tight budget
- Prototyping a video pipeline before scaling up
- Simple looping backgrounds
Other video models
CogVideoX-5B questions
Every figure below is read from the same catalogue entry as the table above, so the two cannot disagree.
What does CogVideoX-5B cost on AI Token Router, and what is the official rate?
CogVideoX-5B is billed at $0.017 per second here, against Zhipu AI's official $0.030 — 43% below, or $0.013 per second back. There is no platform fee, no minimum spend and no per-job surcharge on top of that rate, and failed requests are not billed. The official figure is published next to ours because a discount with no stated benchmark is not a claim anyone can check.
Can I call CogVideoX-5B right now?
No. CogVideoX-5B is catalogued but not currently servable — no configured upstream provider carries it, so a request naming `zhipuai/cogvideox-5b` returns `no available channel` instead of a result. The rate on this page is what it will cost once an upstream serves it, not something you can spend today. No video model in this catalogue is servable yet. What is callable today is Kimi K2.6, Kimi K3, GLM-5.2, DeepSeek V4 Pro, DeepSeek V4 Flash, Qwen-Image and Qwen3 Embedding 8B. 7 of the 24 catalogued models are callable, and every page says which it is.
Why is every CogVideoX-5B generation billed at a fixed duration?
Because a flat price has to be exact rather than approximately right. The billing layer prices a video call as one flat amount per call and does not scale it with a request parameter, so a per-second model routed through it is either exact at a single fixed duration or wrong at every other one. Rather than charge a one-second request what a 6-second request costs, the endpoint accepts `duration` only at 5 seconds and refuses anything else before the request reaches the gateway. At $0.017/sec that makes one CogVideoX-5B clip $0.0850, known before it renders. Variable-length billing is planned; until it ships, the quoted number is the charged number.
How do I generate a clip with CogVideoX-5B?
Video generation is asynchronous rather than a chat call. POST to `https://router.xark.io/api/v1/video/generations` with `"model": "zhipuai/cogvideox-5b"` and a prompt, using the same `Authorization: Bearer` header as every other endpoint, then poll the returned job id for the finished clip. The job response carries `estimated_cost_usd`, so the charge is known before the render starts rather than after it. The bare name `cogvideox-5b`, without the provider prefix, resolves to the same model.
How long a clip can CogVideoX-5B produce, and at what resolution?
Up to 6 seconds per clip, at 480p / 720p. The longest in this catalogue is LTX-2.5 at 20 seconds. Note that the model's maximum and the billed duration are two different numbers during this rollout: generations run at a fixed 5 seconds so the quoted price is exact. Longer output than that comes from stitching clips, which bills as the number of calls you make.
What licence is CogVideoX-5B under, and can I use it commercially?
CogVideoX-5B is published by Zhipu AI under the Apache 2.0 licence. Apache 2.0 is a standard permissive licence: commercial use, modification and redistribution are allowed, subject to keeping the notices and stating what you changed. It also carries an explicit patent grant from the contributors, which terminates if you bring a patent claim over the work — that grant is usually the reason a legal team prefers it to MIT. No user threshold and no field-of-use restriction apply.
How does CogVideoX-5B pricing here compare with other providers carrying it?
Of the 2 other providers we track carrying CogVideoX-5B, the lowest is fal.ai at $0.035/sec, against our $0.017/sec. That is where the rates stood when we last reconciled them, and it can change without notice — we are not claiming to be the cheapest source for this model, only that our rate sits below Zhipu AI's official $0.030/sec. If you find one of these figures stale, tell us and we will correct it.