Concurrency depends on the account’s active package and is shared across API keys. Image tasks consume slots by image count; querying tasks does not consume concurrency.Source for Request limits (opens in a new tab)
Descript UI label: Kling O3 Pro. No confirmed native provider API selector or pinned backend checkpoint is exposed by this source.
Access
Choose this model in Descript App Settings → AI models or the applicable generation tool. Availability depends on the plan; exact native provider checkpoint IDs are not exposed.
Regions
Managed service; exact model-processing region not verified
Application selection label. This access record does not assert that Descript exposes every upstream provider API feature or pins an exact snapshot.
Billing basis
Uses Descript AI credits from a shared plan allowance; exact cost depends on selected model, media length and operation. Provider-native token/character tariffs do not apply to this host.Source for Billing basis (opens in a new tab)
Bundle prices are shared across tools, not a fixed generation price.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Descript help states that production models use no Descript user data; internal research/development uses data only from users who opted in. Enterprise data sharing is disabled by default. Partner agreements prohibit using submitted data to train partner AI. This is not a blanket zero-retention promise; project storage and deletion have separate policies.
Activate the Kling provider in Model Studio, then use a Beijing key. POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis; Authorization: Bearer <key>, X-DashScope-Async: enable, Content-Type: application/json. Poll GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}.
mode=pro (1080p) is default; std=720p; 4k is unavailable for Turbo.
Video duration
Ordinary generation: integer 3–15 seconds, default 5. Feature-video reference: integer 3–10, default 5. Video editing: follows input duration (3–15.5 seconds); duration field ignored.Source for Video duration (opens in a new tab)
Explicit aspect_ratio required for text/reference generation. First-frame tasks inherit that image ratio; editing inherits input video ratio.
Reference images
First frame: 1. First+last: 2. Image-only reference: at most 7 combined images/subjects. With video: at most 4 combined images/subjects, or one first frame.Source for Reference images (opens in a new tab)
Multi-shot generation
multi_shot=false by default. When true, shot_type is required: intelligence or customize. Custom mode uses 1–6 indexed shots, each prompt ≤512 characters (excess is truncated), duration integer 1–parameters.duration.Source for Multi-shot generation (opens in a new tab)
Feature-video reference supports intelligence only. Video editing does not support multi-shot.
Audio generation
audio=false by default. Can generate sound without input video. With feature/base video, audio must remain false; keep_original_sound may preserve the source sound.Source for Audio generation (opens in a new tab)
Reusable subjects
First-frame and first+last-frame generation: up to three subject entities. Image reference: up to seven combined images/entities; with input video: up to four. Video-type entity: at most one, excludes images and image-type entities.Source for Reusable subjects (opens in a new tab)
These are input-video restrictions, not an output FPS promise.
Video reference combinations
feature alone; feature+refer (≤4 images/subjects); feature+first_frame; refer alone (≤7 images/subjects). Editing uses base alone or base+refer (≤4 images/subjects). A video-type entity excludes image references/entities.Source for Video reference combinations (opens in a new tab)
For base-video editing, billed duration is input duration rounded to nearest integer; half-second tie behavior is not specified. Other generation uses output seconds. usage.duration is the billing total.Source for Output billing (opens in a new tab)
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 14 Oct 2026
Category
Usage
Rate
Unit
Video generation
720p · silent · no input video1
CN¥0.60
second of output video
Video generation
720p · generated audio · no input video2
CN¥0.90
second of output video
Video generation
1080p · silent · no input video1
CN¥0.80
second of output video
Video generation
1080p · generated audio · no input video2
CN¥1.20
second of output video
Video generation
4K · silent · no input video1
CN¥3.00
second of output video
Video generation
4K · generated audio · no input video2
CN¥3.00
second of output video
Video generation
720p · input video · no generated audio3
CN¥0.90
second of output video
Video generation
1080p · input video · no generated audio3
CN¥1.20
second of output video
Video generation
4K · input video · no generated audio3
CN¥3.00
second of output video
Rate conditions
Original Beijing CNY price; promotions/taxes excluded. No free quota. audio=false; no base/feature input video. Applies only to kling/kling-v3-omni-video-generation through Alibaba, not the native Kling API.
Original Beijing CNY price; promotions/taxes excluded. No free quota. audio=true; no base/feature input video. Applies only to kling/kling-v3-omni-video-generation through Alibaba, not the native Kling API.
Original Beijing CNY price; promotions/taxes excluded. No free quota. Feature reference or base editing with input video. audio must be false; preserved source sound is a separate option. Editing duration is rounded to nearest integer per API; half-second ties are unspecified. The 720p/1080p card labels omit “No Audio”, but the API forbids generated audio with video input. Applies only to kling/kling-v3-omni-video-generation through Alibaba, not the native Kling API.
Kling AI is the inference provider; Alibaba supplies this regional access contract. Hosted input retention, training use and inference data residency are not independently established. Generated URLs expire after 30 days.
Caching
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
Kling AI API
kling-v3-omni
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Descript editor
Not verified
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How it works
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Minimum reusable input
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Lifetime & refresh
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How to enable it
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Usage fields
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Billing rules
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Limits & data boundaries
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Inference from the exact named production model or managed service in Kuaishou’s official API catalog: no downloadable production weights or open-weight license were established for this identity. Original public-weight Kolors and research derivatives are excluded; public demo code is not evidence of production weights.
