Models

You can read the model catalog without any key. GET /v1/models and GET /v1/models/{id} are public and give you model ids, today's prices and, for media models, the input schema that /v1/generate checks against. The models page on the site is built from the same data.

Listing the catalog

http

GET https://kadraris.com/v1/models

Query parameter	Allowed values	Effect
`category`	`image`, `video`, `audio`, `chat`, `tool`	Return only one category
`currency`	`USD`, `PLN`	Show prices in this currency

bash

curl "https://kadraris.com/v1/models?category=video&currency=USD"

python

import requests

models = requests.get(
    "https://kadraris.com/v1/models",
    params={"category": "video", "currency": "USD"},
    timeout=30,
).json()["data"]

for m in models:
    print(m["id"], m["pricing"]["display"])

javascript

const res = await fetch("https://kadraris.com/v1/models?category=video&currency=USD");
const { data } = await res.json();
for (const m of data) console.log(m.id, m.pricing.display);

A trimmed response follows. The amounts were valid on October 2, 2026; a live call shows the current ones.

json

{
  "object": "list",
  "data": [
    {
      "id": "seedance-2.5",
      "object": "model",
      "name": "Seedance 2.5",
      "vendor": "ByteDance",
      "category": "video",
      "modes": ["text-to-video", "image-to-video", "first-last-frame", "reference-to-video", "video-edit", "video-extend"],
      "pricing": {
        "unit": "second",
        "amount": "0.58701",
        "currency": "USD",
        "display": "$0.587 / second",
        "variants": [
          { "resolution": "480p", "amount": "0.2609" },
          { "resolution": "720p", "amount": "0.58701" },
          { "resolution": "1080p", "amount": "1.01608" }
        ]
      }
    }
  ]
}

Schemas are left out of the list to keep it small. Request a single model with GET /v1/models/{id} to get input_schema.

Fields of a model

Field	Type	Content
`id`	string	The value to send as `model`; it does not change
`name`	string	Human-readable name
`vendor`	string	The company behind the model
`category`	string	`image`, `video`, `audio`, `chat` or `tool`
`object`	string	Always the string `model`
`modes`	array	What the model can do (see below)
`pricing`	object	Price data; its shape differs between media and chat
`input_schema`	object	Accepted `input` parameters; media models only, single-model endpoint only
`context_window`, `max_output_tokens`, `anthropic_compatible`	number, number, boolean	Present on chat models

Reading `modes`

modes tells you which jobs a model handles, which helps you shortlist models before looking at their schemas. Values you may meet:

text-to-image, text-to-video - output from a text prompt only;
image-to-image, image-edit, inpainting - changes to an image sent in image_urls;
image-to-video - a still image, sent as image_url, becomes the opening frame of a clip;
first-last-frame - a clip that moves from one given frame to another;
reference-to-video, multi-image-reference - output guided by reference images or clips;
video-edit, video-extend - changing or lengthening a clip you supply;
lyrics-to-music - a song built from lyrics;
stem-separation, vocal-isolation - splitting audio into separate tracks;
chat - a conversational model.

You never send a mode in the request. The mode follows from the input fields you fill in: for example, a model with image-to-video animates a picture once you pass image_url.

`pricing` on media models

Field	Content
`unit`	The thing you pay for, such as one image or one second of video
`amount`	Unit price as a decimal string
`currency`	`USD` or `PLN`
`display`	Formatted text you can show to users, e.g. `$0.587 / second`
`variants`	Unit price for each value of the parameter that changes the price, often `resolution`; empty if there is only one price
`vat_included`	`true` on PLN prices, which contain 23% VAT; missing on USD

`pricing` on chat models

Field	Content
`input_per_mtok`	Cost of one million input tokens
`output_per_mtok`	Cost of one million output tokens
`currency`	`USD` or `PLN`
`vat_included`	`true` on PLN prices

PLN prices always include 23% VAT. A few current examples: GPT Image 2.5 $0.05 per image, FLUX.2 [pro] from $0.0688 per image, Veo 3.1 (Quality) from $2.53 per video, Claude Opus 5.5 $3.78 / $18.90 per 1M tokens.

One model and its schema

bash

curl https://kadraris.com/v1/models/seedance-2.5

python

model = requests.get("https://kadraris.com/v1/models/seedance-2.5", timeout=30).json()
print(model["input_schema"])

javascript

const model = await (await fetch("https://kadraris.com/v1/models/seedance-2.5")).json();
console.log(model.input_schema);

Ask for an id that does not exist and you get 404 not_found.

How `input_schema` works

The schema lists every parameter /v1/generate will take inside input, written in the style of JSON Schema. Here is a shortened example; always check the live schema for actual values:

json

{
  "type": "object",
  "properties": {
    "prompt":         { "type": "string" },
    "duration":       { "type": "integer", "minimum": 4, "maximum": 10, "default": 5 },
    "resolution":     { "type": "string", "enum": ["480p", "720p"], "default": "720p" },
    "aspect_ratio":   { "type": "string", "enum": ["16:9", "9:16", "1:1"], "default": "16:9" },
    "image_url":      { "type": "string" },
    "generate_audio": { "type": "boolean", "default": false }
  },
  "required": ["prompt"]
}

Keyword	What it tells you
`type`	`string`, `integer`, `number`, `boolean` or `array`
`enum`	The complete set of allowed values
`minimum`, `maximum`	Lowest and highest allowed number, both included
`default`	What the server uses if you skip the field
`required`	Fields you must always send

The server does not tolerate anything outside the schema. Extra fields, wrong types, values not in enum and numbers out of range all produce 422 invalid_request. In that case no task exists and nothing is reserved on your balance.

Validating on your side first

python

def build_input(schema: dict, **values) -> dict:
    props = schema.get("properties", {})
    unknown = set(values) - set(props)
    if unknown:
        raise ValueError(f"unknown fields: {unknown}")
    missing = [f for f in schema.get("required", []) if f not in values]
    if missing:
        raise ValueError(f"missing fields: {missing}")
    for name, value in values.items():
        rule = props[name]
        if "enum" in rule and value not in rule["enum"]:
            raise ValueError(f"{name} must be one of {rule['enum']}")
        if "minimum" in rule and value < rule["minimum"]:
            raise ValueError(f"{name} must be >= {rule['minimum']}")
        if "maximum" in rule and value > rule["maximum"]:
            raise ValueError(f"{name} must be <= {rule['maximum']}")
    return values

schema = requests.get("https://kadraris.com/v1/models/seedance-2.5", timeout=30).json()["input_schema"]
payload = {"model": "seedance-2.5", "input": build_input(schema, prompt="Fog over a harbor", duration=5)}

The API will run the same checks regardless. Doing them in your app lets you show users a clear message before any request is sent.

Working out the cost in advance

Media: unit price multiplied by the number of units, for instance seconds of video times the per-second rate. /v1/generate returns the precise number as estimated_cost and reserves it until the task ends.
Chat: take the token counts from usage, multiply them by input_per_mtok and output_per_mtok, and divide by 1,000,000.

See Billing for the rest.

Models

Listing the catalog

Fields of a model

Reading modes

pricing on media models

pricing on chat models

One model and its schema

How input_schema works

Validating on your side first

Working out the cost in advance

Reading `modes`

`pricing` on media models

`pricing` on chat models

How `input_schema` works