Estimate a generation's cost: POST /v1/estimate

POST /v1/estimate prices a Percify generation in credits before you run it, including per-second lip-sync and quality multipliers. It never charges.

POST https://api.percify.io/v3/playground/v1/estimate returns the exact number of credits a generation will cost, without running it or charging you. It takes the same modelId and input as POST /v1/run and uses the same pricing, so the number matches what the run charges.

Use it for any model whose price depends on its inputs: lip-sync billed per second of audio, voice cloning billed by the length of the voice sample, and image or video models where quality, resolution or duration change the price.

Body

modelId · string · required

A model id from GET /v1/models.

input · object · required

The inputs you plan to send to /v1/run. Required fields must be present and enum values must be valid. For duration-billed models, include the audio or video URL so Percify can measure it.

Response

data.modelId · string

The model that was priced.

data.credits · integer

The credits /v1/run will charge for this input.

data.pricingNote · string

How the model is billed, in words: for example flat 2 credits per run, billed per second of audio, or a note that names the multipliers for a resolution or quality field.

Example

cURL

curl -s -X POST https://api.percify.io/v3/playground/v1/estimate \
  -H "Authorization: Bearer $PERCIFY_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"modelId":"flux-schnell","input":{"prompt":"a lighthouse at dawn, watercolor"}}'

201 Response (flux-schnell)

{
  "success": true,
  "data": {
    "modelId": "flux-schnell",
    "credits": 2,
    "pricingNote": "flat 2 credits per run"
  }
}

How is a duration-billed model priced?

Percify downloads the start of the audio or video you pass and reads its length. If it cannot read the length, the estimate assumes 30 seconds, while /v1/run may refuse the same file (for example an MP4 without a readable duration). If the estimate looks like a 30-second price for a clip of a different length, check that the URL is public and the file is a standard .mp3, .wav or .mp4.

How is it different from a run?

	/v1/estimate	/v1/run
Charges credits	No	Yes
Checks your balance	No	Yes
Content moderation	No	Yes
Starts a generation	No	Yes
Counts toward 60 requests a minute per key	Yes	Yes

A good estimate does not guarantee the run starts: the run can still stop on moderation, your balance or a key’s Monthly credit cap.

Errors

Status	Message starts with	Cause
400	Missing required field: …	Add the field.
400	<field> must be one of: …	Use a listed value.
404	Model '…' not found	Use an id from GET /v1/models.

Related

Start a generation

Run it once the price is right.

List models

Each model’s pricing type and inputs.

Check credits and usage

Balance, monthly spend and key caps.

Talking avatar video

Lip-sync pricing per second of audio.