AI生成 into the corner of the output under China’s labelling rules — NovAI cannot switch it off. Video output varies by model: in our 4 August 2026 tests Seedance 2.0 and CogVideoX-Flash showed no visible label, while Hunyuan Video 1.5 is labelled. Text and chat models are not affected. Which models, and what it rules out →We run NovAI, so this is a first-party guide, not an independent review. Every price below comes from the provider's published pricing page, checked July 2026.
Search "free video generation API" and you get three very different things:
| Type | What it really means | Examples |
|---|---|---|
| Trial credits | $5–$10 on signup, then you pay. Often card required. | Most hosted API platforms |
| Free web app | Daily free generations in a browser UI — but no scriptable API. | Consumer video tools, AI studio web UIs |
| Truly free API | $0 per call, forever, callable from code, no card. | cogvideox-flash (this post) |
| Provider / model | Published price | Free API tier? |
|---|---|---|
| fal.ai — Seedance 2.0 (1080p) | ≈ $0.685 / second | No |
| fal.ai — Seedance V1 Pro Fast (1080p, 5s) | ≈ $0.245 / clip | No |
| fal.ai — Kling 2.1 Standard (5s) | $0.25 / clip | No |
| poyo.ai — Seedance 2.0 (1080p) | ≈ $0.45 / second | No |
| Kling developer portal | from ≈ $0.112 / second | No |
| Z.ai international — CogVideoX-3 / Vidu | paid, per generation | No |
| NovAI — cogvideox-flash | $0 / generation | Yes — no card |
Sources: fal.ai model pages, poyo.ai pricing comparison, kling.ai/dev/pricing, docs.z.ai — all as published July 2026. Prices change; check the linked providers for current numbers.
Zhipu (Z.ai) offers its "flash" models free on its China-side platform. The international Z.ai API only lists paid video models (CogVideoX-3, Vidu). NovAI bridges that gap: we pass the free tier through to international developers with zero markup, email signup, no Chinese phone number.
import requests, time
BASE = "https://aiapi-pro.com/v1"
KEY = {"Authorization": "Bearer sk-novai-xxx"}
# 1. Submit the generation task (async)
task = requests.post(f"{BASE}/video/generations", headers=KEY, json={
"model": "cogvideox-flash",
"prompt": "a paper plane gliding over a neon city at night",
}).json()
print(task) # {"id": "...", "status": "queued", ...}
# 2. Poll until done
while True:
r = requests.get(f"{BASE}/video/generations/{task['id']}", headers=KEY).json()
if r.get("status") not in ("queued", "processing"):
break
time.sleep(5)
print(r) # contains the video URL when finishedThe same API key and the same /v1/video/generations endpoint also serve paid models:
| Model | Price (NovAI) |
|---|---|
hy-video-1.5 (Tencent Hunyuan) | $0.21 / generation |
doubao-seedance-2.0-fast | $0.054–$0.802 / s (480p–4K) |
doubao-seedance-2.0 | $0.067–$1.00 / s (480p–4K); 1080p = $0.329/s |
Full per-second comparison against fal.ai and poyo.ai: AI Video API Pricing 2026.
Email signup, grab a key, generate your first video at $0.
Correct. You register with email and can call cogvideox-flash (and 4 other free models, including image generation) without adding any payment method.
Light workloads, yes. For volume or higher quality, switch the model ID to a paid one — nothing else in your code changes.
Same deal with cogview-3-flash — see Free AI Image Generation API in 2026.