Video API

Video generation

Asynchronous API for Text-to-Video and Image-to-Video generation powered by Kling AI, WanX, MiniMax, Grok, and Seedance. Submit tasks, poll for completion, and download MP4.

Models & capabilities

Supported models, resolutions, durations, audio support, and pricing.

Model & Vendor	Mode	Duration	Resolution & Format	Audio / Sound	References	Pricing
`kling-3.0`Kling AI	Text-to-Video Image-to-Video	3–15 s	720p, 1080p16:9, 9:16 (до 1080p)	Native AI audio `settings.audio: "native"`	Up to 2 imagesfirst_frame, last_frame	$0.14 / s
`kling-3.0-omni`Kling AI	T2V, Multi-ref, I2V	3–15 s	720p, 1080p16:9, 9:16 (до 1080p)	Native AI audio `settings.audio: "native"`	Up to 4 imagesfirst/last + 3 refs	$0.145 / s
`wan3.0-t2v`Alibaba (WanX)	Text-to-Video	2–30 s	480p, 720p, 1080p16:9, 9:16	Silent	—	$0.038 / s
`wan3.0-i2v`Alibaba (WanX)	Image-to-Video	2–30 s	480p, 720p, 1080p16:9, 9:16	Silent	1 imagefirst_frame (start)	$0.038 / s
`wan3.0-prime-t2v`Alibaba (WanX Prime)	Prime Text-to-Video	2–30 s	480p, 720p, 1080p16:9, 9:16	Silent	—	$0.044 / s
`wan3.0-prime-i2v`Alibaba (WanX Prime)	Prime Image-to-Video	2–30 s	480p, 720p, 1080p16:9, 9:16	Silent	1 imagefirst_frame (start)	$0.044 / s
`minimax_h3`MiniMax (Hailuo H3)	Мультимодальный T2V/I2V	4–15 s(default 5)	480p, 768p, 1080p16:9, 9:16, 1:1, 3:4, 4:3, 21:9	Audio ref `type: "audio"`	Up to 9 img, 3 vid, 3 audio	$0.024 / s
`grok-imagine-video-1.5`xAI (Grok)	Text-to-Video, I2V	6–15 s(default 6)	480p, 720p16:9, 9:16, 1:1	Silent	Up to 7 imagesfirst/last + refs	$0.180 / gen
`seedance2.5-01-720`ByteDance (Seedance)	Длинное видео (Text/Multi-ref)	30 s(fixed)	720p16:9, 9:16, 1:1, 3:4, 4:3, 21:9	Silent	Up to 9 images	$0.480 / gen

Task lifecycle

Video generation is performed asynchronously in three steps.

Task Submission (POST /v1/videos): validates parameters, calculates cost, and reserves the balance. Returns a JSON payload with id (e.g. task_01jvideo_example), status queued, poll_url, and content_url.
Status Polling (GET /v1/videos/{task_id}): the client polls every 10–15 seconds. Status transitions: queued → in_progress → completed or failed. Once completed, funds are settled. If the task fails, the balance hold is released automatically with zero charge.
Video Streaming & Download (GET /v1/videos/{task_id}/content): streams the completed MP4 video with full HTTP Range support (status 206 Partial Content) for streaming video players.

Sound & audio settings

How to configure native AI sound synthesis or pass audio reference tracks.

Kling AI (kling-3.0, kling-3.0-omni): supports native AI sound synthesis (engine sounds, ocean waves, footsteps, ambient noise) synchronized with visual motion. To generate video with sound, include:
"settings": {"audio": "native"}.
By default, video is silent ("settings": {"audio": "off"}). Multi-shot camera cuts are also supported: "settings": {"multi_shot": true}.
MiniMax Hailuo H3 (minimax_h3): supports audio reference conditioning. Provide a soundtrack or voice track in references with type: "audio":
"references": [{"type": "audio", "url": "https://example.com/track.mp3"}]
(Up to 3 audio files, 2–15 seconds duration, up to 15 MB).
Alibaba WanX, ByteDance Seedance, xAI Grok: generate silent MP4 video tracks. When audio is required, choose Kling 3.0 with native audio or post-process with an audio layer.

Request parameters

JSON body fields for POST /v1/videos.

Parameter	Type	Description
`model`	string	Required. Video model ID from the catalog.
`prompt`	string	Text description of the scene, subjects, and camera movements (required for Text-to-Video).
`seconds`	integer	Video duration in seconds. Kling: 3–15; WanX: 2–30; MiniMax: 4–15 (default 5); Grok: 6–15 (default 6); Seedance: 30 (fixed).
`size` / `resolution`	string	Video resolution: `480p`, `720p`, `768p`, `1080p`, or exact dimensions (`1280x720`, `1920x1080`).
`aspect_ratio` / `ratio`	string	Aspect ratio: `16:9` (landscape), `9:16` (portrait/Reels/Shorts), `1:1`, `3:4`, `4:3`, `21:9`.
`settings`	object	Kling AI settings: `audio` (`"native"` for AI sound, or `"off"`) and `multi_shot` (boolean).
`references`	array	Array of reference objects with `type` (`"image"`, `"audio"`), `role` (`"first_frame"`, `"last_frame"`, `"reference"`), and `url`.
`prompt_extend`	boolean	For WanX models: automatic prompt expansion by upstream LLM (default `false`).

Examples

Ready-to-use examples for cURL and Python.

bash

curl https://api.gateyourway.com/v1/videos \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer gyw-sk-YOUR_KEY" \
  -d '{
    "model": "kling-3.0",
    "prompt": "A cinematic aerial shot of ocean waves crashing on dark rocks at sunset with seagulls flying",
    "seconds": 5,
    "size": "1920x1080",
    "settings": {
      "audio": "native",
      "multi_shot": false
    }
  }'

json

{
  "id": "task_01jvideo_example",
  "status": "queued",
  "model": "kling-3.0",
  "seconds": 5,
  "poll_url": "/v1/videos/task_01jvideo_example",
  "content_url": "/v1/videos/task_01jvideo_example/content"
}

bash

curl https://api.gateyourway.com/v1/videos \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer gyw-sk-YOUR_KEY" \
  -d '{
    "model": "wan3.0-i2v",
    "prompt": "The camera slowly zooms in as gentle wind blows through cherry blossom petals",
    "seconds": 5,
    "size": "720p",
    "ratio": "16:9",
    "prompt_extend": false,
    "references": [
      {
        "type": "image",
        "role": "first_frame",
        "url": "https://example.com/start_frame.jpg"
      }
    ]
  }'

bash

curl https://api.gateyourway.com/v1/videos \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer gyw-sk-YOUR_KEY" \
  -d '{
    "model": "minimax_h3",
    "prompt": "A cyberpunk street market in neo-Tokyo with rain reflections and neon signs",
    "seconds": 6,
    "size": "768p",
    "aspect_ratio": "16:9",
    "references": [
      {
        "type": "audio",
        "url": "https://example.com/synthwave_bgm.mp3"
      }
    ]
  }'

bash

# 1. Проверка статуса (поллинг раз в 10-15 секунд):
curl https://api.gateyourway.com/v1/videos/task_01jvideo_example \
  -H "Authorization: Bearer gyw-sk-YOUR_KEY"

# Ответ при готовности:
# {
#   "id": "task_01jvideo_example",
#   "status": "completed",
#   "content_url": "/v1/videos/task_01jvideo_example/content"
# }

# 2. Скачивание готового MP4:
curl -L https://api.gateyourway.com/v1/videos/task_01jvideo_example/content \
  -H "Authorization: Bearer gyw-sk-YOUR_KEY" \
  --output video.mp4

python

import os
from pathlib import Path
import time
import requests

api_base = "https://api.gateyourway.com/v1"
user_key = os.getenv("GATEYOURWAY_API_KEY", "gyw-sk-YOUR_KEY")
req_headers = {
    "Authorization": f"Bearer {user_key}",
    "Content-Type": "application/json",
}

# 1. Запуск генерации видео с родным звуком (Kling 3.0)
payload = {
    "model": "kling-3.0",
    "prompt": "A vintage sports car driving fast along a scenic coastal highway at sunset",
    "seconds": 5,
    "size": "1920x1080",
    "settings": {
        "audio": "native",
        "multi_shot": False,
    },
}

resp = requests.post(f"{api_base}/videos", json=payload, headers=req_headers)
resp.raise_for_status()
task = resp.json()
task_id = str(task["id"])
print(f"Task created: {task_id}, status: {task.get('status')}")

# 2. Опрос статуса задачи
while True:
    time.sleep(10)
    poll_resp = requests.get(f"{api_base}/videos/{task_id}", headers=req_headers)
    poll_resp.raise_for_status()
    data = poll_resp.json()
    status = data.get("status")
    print(f"Current status: {status}")
    if status == "completed":
        break
    elif status == "failed":
        raise RuntimeError(f"Generation failed: {data.get('error')}")

# 3. Скачивание готового MP4 файла
print("Downloading MP4 video...")
video_resp = requests.get(f"{api_base}/videos/{task_id}/content", headers=req_headers)
video_resp.raise_for_status()
Path("video.mp4").write_bytes(video_resp.content)
print("Saved to video.mp4 successfully!")