Video models are quoted in at least three different units — per second, per generation, per 1M tokens — which makes "which is cheaper" almost impossible to answer from a pricing page. This post converts all of them to the same unit: one 5-second clip at the model's default resolution.
| Model | Provider | Per second | Per 5s clip | Billing |
|---|---|---|---|---|
| PixVerse C1 | PixVerse | $0.0308 | $0.15 | token |
| PixVerse V5.6 / V6.0 | PixVerse | $0.0406 | $0.20 | token |
| Hunyuan Video 1.5 | Tencent | $0.0462 | $0.23 | token |
| Seedance 2.0 Fast (480p) | ByteDance | $0.0540 | $0.27 | per second |
| Kling 2.5 Turbo | Kuaishou | $0.0615 | $0.31 | token |
| Hailuo 2.3 / 2.3 Fast | MiniMax | $0.0615 | $0.31 | token |
| Seedance 2.0 (480p) | ByteDance | $0.0670 | $0.34 | per second |
| Kling 2.6 | Kuaishou | $0.0769 | $0.38 | token |
| Hailuo H3 / H3 Max | MiniMax | $0.0769 | $0.38 | token |
| Vidu Q3 | Vidu | $0.0769 | $0.38 | token |
| Kling O1 / V3 / V3 Omni | Kuaishou | $0.0923 | $0.46 | token |
| Vidu Q3 Pro | Vidu | $0.0923 | $0.46 | token |
| Kling V3 Turbo | Kuaishou | $0.1231 | $0.62 | token |
| Seedance 2.0 Fast (720p) | ByteDance | $0.1310 | $0.66 | per second |
| Seedance 2.5 (480p) | ByteDance | $0.1400 | $0.70 | per second |
| Seedance 2.0 (720p) | ByteDance | $0.1630 | $0.82 | per second |
| Seedance 2.0 Fast (1080p) | ByteDance | $0.2640 | $1.32 | per second |
| Seedance 2.5 (720p) | ByteDance | $0.3200 | $1.60 | per second |
| Seedance 2.0 (1080p) | ByteDance | $0.3290 | $1.65 | per second |
| Seedance 2.0 Fast (4K) | ByteDance | $0.8020 | $4.01 | per second |
| Seedance 2.0 (4K) | ByteDance | $1.0000 | $5.00 | per second |
Seedance 2.0 is the clearest example: the same model costs $0.067/s at 480p, $0.163/s at 720p, $0.329/s at 1080p and $1.00/s at 4K — an 18x range inside one model. If you are prototyping, 480p drafts cost a fifth of 720p finals.
Nearly every family ships a cheaper tier: Seedance 2.0 Fast is ~20% cheaper than 2.0 at the same resolution, and Kling 2.5 Turbo ($0.0615/s) is half the price of Kling V3 Turbo ($0.1231/s). For storyboards and iteration, the turbo tier is usually enough.
Hailuo 2.3 and Hailuo H3 are both token-billed, but H3 costs 25% more per second. The premium buys consistency and prompt adherence rather than pixels.
| Workload | Cheapest tier | Mid tier | Flagship / 4K |
|---|---|---|---|
| 100 × 5s clips | $15 (PixVerse C1) | $46 (Kling V3) | $500 (Seedance 2.0 4K) |
| 1,000 × 5s clips | $154 | $461 | $5,000 |
| 1 minute of finished video (12 clips) | $1.85 | $5.5 | $60 |
At these prices the model choice matters far more than the gateway: switching from 4K Seedance 2.0 to 720p Seedance 2.5 cuts a 1,000-clip batch from $5,000 to $1,600, and switching to a token-billed mid tier cuts it to under $400.
CogVideoX-Flash is free to call on NovAI (10 free-model calls per day, 5 free video generations per day on the free tier). It is not a Seedance-class model, but it is enough to wire up the request/queue/download path before you spend anything.
curl https://aiapi-pro.com/v1/video/generations \
-H "Authorization: Bearer $NOVAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"cogvideox-flash","prompt":"a paper boat in the rain, close up"}'Three things drive it: resolution (Seedance 2.0 costs 18x more at 4K than at 480p), model tier (a turbo/fast variant is 2-3x cheaper than the flagship), and billing unit (per-second models quote a flat rate, token-billed models scale with the tokens the run consumes).
They are the providers' list prices passed through with a 1.0 multiplier and zero platform markup. NovAI publishes the feed at /china-ai-leaderboard.json so the numbers can be checked.
Kling, Hailuo, Vidu and PixVerse bill per token. Each model has a token price and an estimated token count per second of output; multiplying them gives the per-second cost, which is what the table shows as an estimate.
Yes. CogVideoX-Flash is free to call with a NovAI API key, capped at 10 free-model calls per day and 5 free video generations per day on the free tier.
111 live models, provider list price, 0% platform markup. $2 credit on signup, no card required.
Get an API keyPrices reflect provider rate cards as of 2026-10-22 and are published machine-readably at /china-ai-leaderboard.json. Token-billed figures are estimates; per-second figures are the published rates.