Video

Use POST /v1/video/generations for synchronous video generation. The HTTP connection stays open while the provider renders the result.

POST https://api.rout.my/v1/video/generations

Request

curl --max-time 900 https://api.rout.my/v1/video/generations \
  -H "Authorization: Bearer $ROUTMY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "provider/video-model-id",
    "prompt": "A slow camera move across a glass cube on a white table",
    "duration": 5,
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "n": 1
  }'

Request fields

Field	Type	Required	Notes
model	string	Yes	Exact video model ID from /v1/models.
prompt	string or object	Yes	Text prompt, or { "text": "...", "image": "..." } for image-to-video.
input_image	string	No	URL, data URI, or raw base64 image for image-to-video.
input_image_b64	string	No	Explicit base64 image input.
image, image_url, input_image_url	string	No	Accepted aliases for image input.
input_video	string	No	URL, data URI, or raw base64 source video for edit mode.
input_video_b64	string	No	Explicit base64 source video input.
input_audio	string	No	URL, data URI, or raw base64 source audio for models that support an external audio track.
input_audio_b64	string	No	Explicit base64 audio input.
duration	number	No	Requested duration in seconds. Maximum is clamped server-side.
resolution	string	No	Values such as 480p, 720p, 1080p, 4K, or WIDTHxHEIGHT.
aspect_ratio, aspectRatio	string	No	Values such as 16:9, 9:16, 1:1, 4:3, 3:4.
n, sampleCount	integer	No	Number of output videos. Defaults to 1.
audio	boolean	No	Explicit audio on/off.
sound	string	No	Audio switch for providers that expose string flags, for example Kling on or off.
tier	string	No	Friendly video quality tier. For Kling, standard maps to mode:"std", pro maps to mode:"pro", and 4k maps to mode:"4k" when supported.
quality, mode	string	No	Values such as draft, std, pro, 4k, lossless, optimized, single, or multi, when supported by the selected model.
negative_prompt, negativePrompt	string	No	Content to avoid.
seed	integer	No	Deterministic generation seed when supported.
last_frame_image, lastFrameImage, image_tail, imageTail	string	No	Ending frame for first/last-frame models.
reference_urls, referenceUrls	string[]	No	Reference media URLs for reference-to-video models. Data URI/base64 entries are temporarily hosted when possible.
reference_images, referenceImages	string[]	No	Reference image URLs for multi-reference models.
reference_videos, referenceVideos	string[]	No	Reference video URLs for models that support video references.
watermark	boolean	No	Watermark flag when supported.
prompt_enhancement, enhance_prompt, prompt_extend, enable_prompt_expansion	boolean	No	Unified prompt expansion/enhancement switch. CamelCase aliases are accepted.
camera_fixed, cameraFixed	boolean	No	Fixed camera flag for supported Seedance models.
cfg_scale, cfgScale	number	No	Guidance scale for supported Kling models.
camera_control, cameraControl	string	No	Kling camera control mode.
camera_control_config, cameraControlConfig	object	No	Kling camera control config object.
shot_type, shotType	string	No	Wan reference-to-video shot type, usually single or multi.
character_orientation, characterOrientation	string	No	Kling motion-control orientation, usually image or video.
keep_original_sound, keepOriginalSound	boolean	No	Kling motion-control source audio flag.
voice_list, voiceList	object[]	No	Kling voice-control voices, up to 2 entries such as { "voice_id": "..." }. Use prompt markers like <<<voice_1>>> and enable audio with sound:"on" or audio:true.
person_generation, personGeneration	string	No	Veo person generation mode when supported.
resize_mode, resizeMode	string	No	Veo resize mode when supported.
compression_quality, compressionQuality	string	No	Veo compression quality when supported.
service_tier, serviceTier	string	No	Service tier for models that support it, for example Seedance default or flex.
audio_url, audioUrl	string	No	Accepted aliases for input_audio.
poll_timeout_ms, pollTimeoutMs	integer	No	Video job polling timeout in milliseconds.
poll_interval_ms, pollIntervalMs	integer	No	Video job polling interval in milliseconds.
user	string	No	Optional client-side user identifier.

Use the exact model IDs returned by /v1/models.

Optional video controls

These fields are accepted at the top level. Availability depends on the selected model.

Field	Notes
audio	Enables model audio when supported.
sound	Kling audio switch, usually on or off.
negative_prompt, negativePrompt	Content to avoid when supported.
prompt_enhancement, enhance_prompt, prompt_extend, enable_prompt_expansion	Prompt expansion/enhancement when supported.
tier, quality, mode	Quality or mode selection, for example Kling std/pro/4k or Wan single/multi.
watermark	Watermark flag when supported.
last_frame_image, lastFrameImage, image_tail, imageTail	Ending frame for first/last-frame models.
voice_list, voiceList	Kling voice control. If prompt uses <<<voice_1>>>, pass matching voices and enable audio.
n, sampleCount	Multiple outputs for models that support them.
reference_urls, referenceUrls	Reference media for -r2v models such as Wan reference-to-video.
reference_images, referenceImages	Multiple image references for models that support them, including Seedance 2.0 and Wan reference-to-video.
reference_videos, referenceVideos	Multiple video references for models that support them, including Seedance 2.0.
input_video, video_url, videoUrl	Source video for xAI video editing, Kling motion-control, or a single Seedance 2.0 video reference.
input_audio, audio_url, audioUrl	External audio track when supported.
poll_timeout_ms, poll_interval_ms	Longer waits for slow video jobs.

Some editing and motion-control models ignore duration, resolution, and aspect_ratio.

Kling tiers, audio, and voice control

Kling Standard / Pro / 4K can be selected with tier, mode, or quality:

{
  "model": "klingai/kling-v3.0-t2v",
  "prompt": "A chef plates a dessert with cinematic lighting.",
  "tier": "pro",
  "sound": "on",
  "duration": 5,
  "aspect_ratio": "16:9"
}

tier:"standard" becomes Kling mode std, tier:"pro" becomes pro, and tier:"4k" becomes 4k when the selected Kling model supports it.

For voice control, add voice_list and reference voices from the prompt:

{
  "model": "klingai/kling-v3.0-t2v",
  "prompt": "The man <<<voice_1>>> said, \"Hello.\"",
  "tier": "pro",
  "sound": "on",
  "voice_list": [
    { "voice_id": "voice-id-from-provider" }
  ]
}

Voice control accepts up to 2 voices. If the prompt contains <<<voice_1>>> markers, voice_list is required and sound must be enabled.

Veo

Use Veo model IDs from /v1/models, for example google/veo-3.1 or google/veo-3.1-fast.

{
  "model": "google/veo-3.1-fast",
  "prompt": "A robot skating through a neon arcade, cinematic motion",
  "duration": 6,
  "resolution": "720p",
  "aspect_ratio": "16:9",
  "n": 2,
  "audio": true
}

For Veo text-to-video, duration must be 4, 6, or 8; resolution must be 720p or 1080p (4K is also accepted on Veo 3.1 models); aspect_ratio must be 16:9 or 9:16. Use n/sampleCount for multiple outputs when supported.

Image-to-video

{
  "model": "provider/video-model-id",
  "prompt": {
    "text": "Animate this product with a slow turntable movement.",
    "image": "https://example.com/source.png"
  },
  "duration": 5,
  "resolution": "720p"
}

Base64 and data URI images are temporarily hosted as URLs when needed.

xAI Grok Imagine 1.5 Preview

x-ai/grok-imagine-video-1.5-preview is image-to-video. It requires an image input and does not support text-to-video.

{
  "model": "x-ai/grok-imagine-video-1.5-preview",
  "prompt": "Slow cinematic push-in as embers drift across the battlefield and the helmet's crest stirs in the wind",
  "image_url": "https://your-host.com/helmet.jpg",
  "duration": 10,
  "resolution": "720p"
}

Accepted image aliases are input_image, image_url, and prompt:{"image":"..."}. input_video is not used for this model.

Seedance 2.0 Fast

bytedance/seedance-2.0-fast supports text-to-video, image-to-video, and multimodal references through reference_images, reference_videos, input_video, and input_audio.

{
  "model": "bytedance/seedance-2.0-fast",
  "prompt": "Use the image as visual reference and animate it with a slow cinematic camera move.",
  "input_image": "https://example.com/reference.png",
  "duration": 5,
  "resolution": "480p",
  "aspect_ratio": "16:9"
}

For multiple image references:

{
  "model": "bytedance/seedance-2.0-fast",
  "prompt": "Use [Image 1] as the main character and [Image 2] as the product reference in a cinematic street scene.",
  "reference_images": [
    "https://example.com/character.png",
    "https://example.com/product.png"
  ],
  "duration": 5,
  "resolution": "720p",
  "aspect_ratio": "16:9"
}

For a video reference, use input_video for one clip or reference_videos for multiple clips. For an audio reference, use input_audio or audio_url. When input_image is combined with audio/video references on Seedance 2.0, rout.my treats the image as a reference image to avoid first-frame/reference-media conflicts.

{
  "model": "bytedance/seedance-2.0-fast",
  "prompt": "Use the image references for characters, the video reference for motion, and the audio reference for soundtrack timing.",
  "reference_images": [
    "https://example.com/character.png",
    "https://example.com/background.png"
  ],
  "reference_videos": [
    "https://example.com/motion.mp4"
  ],
  "input_audio": "https://example.com/music.mp3",
  "duration": 5,
  "resolution": "480p",
  "aspect_ratio": "16:9"
}

Video editing

{
  "model": "provider/video-model-id",
  "prompt": "Make the camera movement smoother and keep the same subject.",
  "input_video": "https://example.com/source.mp4",
  "duration": 5,
  "resolution": "720p"
}

input_video triggers video-input accounting. Base64 videos can be passed through input_video_b64 and are temporarily hosted when needed.

For some edit and motion-control models, duration, resolution, and aspect_ratio are not supported and may be ignored.

Motion control

{
  "model": "klingai/kling-v2.6-motion-control",
  "prompt": {
    "image": "https://example.com/character.png",
    "text": "Make the character follow the reference motion."
  },
  "input_video": "https://example.com/reference-motion.mp4",
  "mode": "pro",
  "character_orientation": "image",
  "keep_original_sound": false
}

Motion-control models require both a character image and a reference video URL.

Reference-to-video

{
  "model": "qwen/wan-v2.6-r2v",
  "prompt": "character1 and character2 talk in a cafe",
  "reference_urls": [
    "https://example.com/character1.png",
    "https://example.com/character2.png"
  ],
  "duration": 4,
  "resolution": "1080p",
  "mode": "single"
}

reference_urls is required for -r2v models. These models use reference media instead of a primary input_image.

Response

{
  "created": 1744000000,
  "model": "provider/video-model-id",
  "duration": 5,
  "videos": [
    {
      "url": "https://api.rout.my/public/media/generated.mp4",
      "mime_type": "video/mp4"
    }
  ]
}

Generated video URLs are temporary. Download and re-host them if your application needs long-term storage.

Quota accounting

Video requests use a duration-based formula:

tokens = duration_sec x 60,000 x resolution x audio x quality x video_input x n x token_multiplier

Common coefficients:

Factor	Value	Multiplier
resolution	480p	0.7
resolution	720p or 1K	1.0
resolution	1080p or 2K	1.8
resolution	4K	2.5
audio	off	1.0
audio	on	1.3
quality	draft	0.3
quality	std or optimized	1.0
quality	pro or lossless	1.5
video input	absent	1.0
video input	present	1.7

Quota is checked before generation starts. If quota is insufficient, the request returns 429.