Kandinsky 6.0 Pro costs $1.9286 per text-to-video clip and $2.00 per image-to-video clip on HiAPI. Both entries produce a five-second 480p clip with synchronized audio, billed per clip, and $10 covers about five text-to-video runs.
The model is live on HiAPI. Open the model page to generate a clip; new accounts get 200 credits ($0.10), so add balance before the first $1.9286 run.
What does one clip cost?
| Entry | Price per clip | Credits per clip | 100 clips | 1,000 clips |
|---|---|---|---|---|
| Text to video | $1.9286 | about 3,857 | $192.86 | $1,928.60 |
| Image to video | $2.00 | 4,000 | $200 | $2,000 |
2,000 credits convert to $1, and the model page shows both the credit and dollar figures. The entry sets the price; aspect ratio, prompt length, negative prompt, guidance_scale and seed leave it untouched.
A first test costs $10 for about five text-to-video clips, $9.643 in total, or exactly five image-to-video clips. A prompt usually takes several rounds to settle, so budget for ten. Run your first clip on the model page →
The same math scales: $50 covers about 25 text-to-video clips, $100 about 51 and $500 about 259. A monthly run of 20 vertical clips stays under $40.
How does per-clip billing work?
Each successful generation is charged one fixed price. There is no duration setting, so there is nothing to estimate: a clip is five seconds. A 16:9 request comes back at 864×480 and 24fps. Audio is generated with the picture, always on and included in the price, with no separate charge and no off switch.
Failed tasks release their hold, so only finished clips are billed. Poll the returned taskId or set callback.url for the final state, and submit once: a repeated POST creates another billable task. The request fields are in the text-to-video docs.
Duration, resolution and upscaling are not exposed. Five seconds at 480p is the whole menu, which keeps the bill simple: clips × unit price, with no add-ons for audio, aspect ratio or parameters. Image to video is billed the same way, and its first frame carries no separate charge.
The night train below is one text-to-video run: 16:9, 864×480, five seconds, $1.9286. Watch the MP4 → The prompt names the train, the steam, the rising camera and the rail rhythm, and the soundtrack arrives with the picture.
How does it compare with per-second video models?
Kandinsky 6.0 Pro sells a bundle: five seconds, 480p and audio for one price. Kling 3.0, Wan 2.7 and Seedance 2.0 bill by the second, so a five-second clip costs whatever the chosen tier adds up to:
| Model | Billing | Five-second total | Tiers |
|---|---|---|---|
| Kandinsky 6.0 Pro text to video | $1.9286 per clip | $1.9286 | Fixed 480p, audio included |
| Kandinsky 6.0 Pro image to video | $2.00 per clip | $2.00 | Fixed 480p, audio included |
| Kling 3.0 text to video | from $0.10 per second | $0.50 (720p), $0.715 (720p + audio), $0.645 (1080p) | 720p / 1080p / 4K, audio on or off |
| Wan 2.7 text to video | from $0.10 per second | $0.50 (720p), $0.835 (1080p) | Up to 1080p |
| Seedance 2.0 | from $0.136 per second | $0.68 (480p), $1.465 (720p) | 480p / 720p |
At five seconds, the flat price sits above the entry tier of all three per-second models. What it buys is a predictable bill and audio in the same pass; the other totals move with the tier and the length you order, so line up the resolution before comparing.
Keep the tiers in view: Kandinsky 6.0 Pro is fixed at 480p, Kling 3.0 and Wan 2.7 reach 720p and above, and Seedance 2.0 offers a 480p tier at the same resolution. When a placement genuinely needs 720p, price the seconds; when it does not, the flat clip is easier to plan around.
- Fixed five seconds at 480p with sound, budgeted per clip: Kandinsky 6.0 Pro. You know the cost before you submit.
- 720p or 1080p: Kling 3.0 or Wan 2.7, billed on the final seconds, with audio as a separate switch on Kling 3.0.
- 480p at a per-second rate: Seedance 2.0, $0.68 for five seconds, for longer or high-volume material where 480p is enough.
FAQ
Am I charged for failed generations? No. The amount held for a failed task is released, and only finished clips are billed.
Does audio cost extra? No. Audio is generated with the picture, always on, and included in the per-clip price.
Can I generate 10 seconds or 720p? No. Output is fixed at five seconds and 480p. For sharper output, use a per-second model with selectable resolution, such as Kling 3.0 or Wan 2.7.
Do the 200 sign-up credits cover a clip? No. They are worth $0.10, while one text-to-video clip costs $1.9286, so top up before submitting.
Run your first clip
The night train prompt is loaded: open the model page, leave the aspect ratio at 16:9 and submit. The clip costs $1.9286 and arrives at five seconds and 864×480, with steam, rail rhythm and a distant whistle on the same track. If the direction works, rerun the same prompt at 9:16 for a vertical cut; the price stays $1.9286.
Related: Kandinsky 6.0 Pro API guide · image-to-video model page · text-to-video docs
