Video and speech models run as tasks, like images. The difference is the unit: video is charged per second of output and speech per 1,000 characters of input text.
| Model | Unit | Price |
|---|---|---|
video/demo-1 | second | $0.100 / second |
audio/demo-tts-1 | 1K characters | $0.030 / 1K characters |
Video
Inputs
| Field | Type | Default | Notes |
|---|---|---|---|
prompt | string | Required. What happens in the clip | |
duration | integer | 5 | Length in seconds, 1 to 10. This is what you are charged for |
aspect_ratio | string | 16:9 | 16:9, 9:16 or 1:1 |
Create a clip
curl https://api.tovano.ai/v1/tasks \
-H "Authorization: Bearer $TOVANO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "video/demo-1",
"input": { "prompt": "Slow push-in on a cup of coffee, steam rising", "duration": 5, "aspect_ratio": "9:16" },
"webhook_url": "https://example.com/hooks/tovano"
}'
A five-second clip reserves and then charges $0.500. Video takes longer than images, so a webhook is the better way to wait.
The result
{
"status": "completed",
"output": {
"video": {
"url": "https://files.tovano.ai/...",
"width": 144,
"height": 256,
"duration": 5,
"has_audio": false
}
},
"unit": "second",
"units": 5,
"charged_micro": 500000
}
Speech
Inputs
| Field | Type | Default | Notes |
|---|---|---|---|
text | string | Required. Up to 5,000 characters. Its length is what you are charged for | |
voice | string | ava | ava, leo or mia |
Create speech
curl https://api.tovano.ai/v1/tasks \
-H "Authorization: Bearer $TOVANO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "audio/demo-tts-1",
"input": { "text": "Your order has shipped and will arrive on Thursday.", "voice": "leo" }
}'
The result has a WAV file and the character count:
{
"status": "completed",
"output": {
"audio": { "url": "https://files.tovano.ai/...", "content_type": "audio/wav", "duration_ms": 3120 },
"characters": 51
},
"unit": "1k_chars",
"units": 0.051,
"charged_micro": 1530
}
Work out the cost first
Both units are known before you send the request, so the estimate is exact:
const PRICE = { video: 100_000, speech: 30_000 }; // micro-dollars per unit
const videoCost = (seconds: number) => seconds * PRICE.video;
const speechCost = (text: string) => Math.ceil((text.length * PRICE.speech) / 1000);
videoCost(5); // 500000 = $0.50
speechCost("Your order has shipped and will arrive on Thursday."); // 1530 = $0.00153
Read live prices from GET /v1/models rather than hard-coding them, so a price change never surprises your own billing.
Tips
- Split long text. Speech requests take up to 5,000 characters. For longer scripts, split at sentence ends and join the files. The total price is the same.
- Keep clips short while testing. Use
duration: 1while you build. It costs $0.100 per run instead of $0.500. - Store the files. Output URLs work for 24 hours; files are deleted after 30 days.