OmniHuman 1.5 costs $0.193 per second of output video on HiAPI, and 720p and 1080p are the same price. An 8-second clip is $1.544, a 15-second clip is $2.895, and a full 60 seconds is $11.58.
Open the OmniHuman 1.5 model page to run a clip; new accounts get 200 credits on sign-up, and those credits are worth $0.10, so top up before a task that needs $3.86 of balance.
What does OmniHuman 1.5 cost per second?
| Resolution | Price per second | Credits per second |
|---|---|---|
| 720p | $0.193 | 386 |
| 1080p | $0.193 | 386 |
One rate covers both resolutions, and 2,000 credits equal $1. Billing follows the output video: the length comes from your audio, rounds up to a whole second, starts at 1 second and stops at 60.
Settings do not move the number. A prompt of up to 300 characters, up to five masks, a seed and fast mode are all included in the per-second rate.
What does a real clip cost?
| Clip length | Cost |
|---|---|
| 1 second | $0.193 |
| 5 seconds | $0.965 |
| 8 seconds | $1.544 |
| 15 seconds | $2.895 |
| 30 seconds | $5.79 |
| 60 seconds | $11.58 |
Every row is identical at 720p and 1080p, so the table is the full price list. The 60-second row is the ceiling for one task.
The presenter above was driven by a 7.5-second recording and billed as 8 seconds: $1.544. Open the model page to try an 8-second clip →
How far does a budget go?
| Budget | Finished video |
|---|---|
| $10 | about 51.8 seconds |
| $25 | about 129.5 seconds |
| $50 | about 259.1 seconds |
| $100 | about 518.1 seconds |
At $10 that is six 8-second clips for $9.264, which is enough to test a portrait, a different take and a re-recording before committing to a longer piece. Start with one clip on the model page →
Why does every task hold $3.86?
Creating a task reserves 20 seconds of video, $3.86, against the account balance. When the video finishes, the task settles to its real duration: a 5-second result releases the difference, and a 30-second result is charged the extra.
Keep at least $3.86 available and a submission will not stall on balance. The hold is the same at 720p and 1080p, and fast mode, prompts, masks and seeds change neither the hold nor the final bill.
How do you spend less for the same clip?
- Record inside 15 seconds. Output is most consistent there, and the bill stops at the length you actually need.
- Trim before submitting. Dead air at the start and a second take at the end both bill at $0.193 per second. Cut them first.
- Choose resolution by delivery target, not by cost. Both cost the same. 1080p is the default; 720p is enough for feeds that downscale the video anyway.
- Draft in fast mode. It returns sooner at the same price with slightly lower image quality, so use it for review rounds and switch it off for the final render.
- Plan around the 60-second cap. Longer narration becomes several tasks, each billed at the same per-second rate.
How does OmniHuman 1.5 compare on price?
| Model | Price per second | What it does differently |
|---|---|---|
| OmniHuman 1.5 | $0.193 | Needs a recorded audio track; people, pets and anime portraits; masks for group shots; 60 seconds per task |
| HeyGen Avatar IV | $0.15 | Script plus voice, or your own audio up to 600 seconds |
| HeyGen Avatar V | $0.15 | Preset avatars for talking-head video |
| Kling 3.0 Omni image-to-video | $0.10 | Prompt-driven image-to-video with native audio, not lip sync |
Per second, OmniHuman 1.5 runs about 29% above HeyGen Avatar IV. The gap on a finished clip is small: 8 seconds is $1.544 against $1.20, a difference of $0.344; 15 seconds is $2.895 against $2.25; 60 seconds is $11.58 against $9.00.
HeyGen Avatar IV also accepts audio up to 600 seconds, which costs $90.00. OmniHuman 1.5 stops at 60 seconds per task.
Price is only part of the choice. HeyGen models are built for human presenters and can turn a script into speech in the same call. OmniHuman 1.5 needs a recording, but it accepts pets, anime characters and illustrations, and masks let one person speak inside a group frame. Kling 3.0 Omni is a different tool: it generates video from a prompt with its own audio rather than matching a mouth to your track.
FAQ
Does 1080p cost more than 720p? No. Both are $0.193 per second, or 386 credits. 1080p is the default output.
What if my audio is 7.5 seconds? The clip bills as 8 seconds, $1.544. Every task rounds up to a whole second, with a 1-second minimum.
Do prompts, masks, seeds or fast mode add a fee? No. The bill depends on output seconds alone; those inputs are free to use.
How much balance do I need before submitting? At least $3.86 to cover the 20-second hold. Whatever the task does not use returns to the balance when it settles.
Run your first clip
Upload a portrait, add an 8-second recording, leave the resolution at 1080p and submit. That clip costs $1.544, and the task holds $3.86 until it finishes. Open the OmniHuman 1.5 model page and generate it →
Related: OmniHuman 1.5 vs HeyGen Avatar IV · five prompt templates · API getting-started guide
