Kling V3 Omni Video API

Kling V3 Omni is Kuaishou's flagship general-purpose video generation model in the Kling series, unifying long video generation, multimodal references (image/video/character), multi-shot auto-composition, and optional audio generation into a single model interface.

On iCreat, the interface supports two capability codes sharing the same request structure and parameters: kwaivgi/kling-video-o1 (Kling Video O1) and kwaivgi/kling-v3-omni (Kling V3 Omni). Developers choose the appropriate capability code and substitute it into the submission endpoint. Both are called through an asynchronous task queue, supporting Standard (720p), Pro (1080p), and 4K (2160p) resolutions, as well as multi-shot composition, reference video style transfer, first/last frame specification, and other advanced capabilities.

Model Positioning

Kling V3 Omni is positioned as the full-featured video generation interface in the Kling series. Unlike Kling V3 T2V (text-only input) and Kling V3 I2V (image + text only), V3 Omni unifies text, image, video, and character materials into a single request structure, supporting multi-shot auto-composition, reference video style transfer, first/last frame specification, audio generation, negative prompts, and watermark control — suited for video production scenarios requiring complete creative control.

For teams that need to generate video from mixed multimodal materials or require multi-shot narrative, V3 Omni provides a single endpoint covering all capabilities. If you only need quick text-to-video or single-image-to-video generation, V3 T2V or V3 I2V are lighter alternatives.

Core Capabilities

Multimodal Input & Image Placeholders

Supports three multimodal input types: image_list (image materials), element_list (subject library), and video_list (video materials). Prompts can reference images in image_list via <<<image_1>>> placeholders for precise image-to-video control.

Multi-Shot Composition

Set multi_shot: true with shot_type: intelligence to let the model automatically decompose a long narrative into 3-6 shots; or use shot_type: customize with 1-6 manual multi_prompt entries for full per-shot control.

First/Last Frame Specification

Specify first and last frame images in image_list via type: first_frame and type: end_frame for precise start/end frame control.

Reference Video & Style Transfer

Use video_list with refer_type: feature to learn the reference video's color grading, lighting, and cinematography style, or refer_type: base to rewrite the scene while preserving the original subject's motion.

Optional Audio Generation

Set sound: on to generate audio synchronized with the video. Note: when video_list is present in the request, sound can only be set to off.

Three Resolution Tiers

Switch via the mode field: std (720p, lowest cost), pro (1080p, enhanced detail), 4k (2160p, premium delivery).

Negative Prompts

Supports negative_prompt to describe content that should not appear in the video, helping the model avoid unwanted visual elements.

Watermark Control

Control watermark via watermark_info.enabled, suitable for brand content scenarios requiring watermark-free delivery.

Pricing

Total = Unit Price × Output Video Duration

Reference video refers to video materials provided via video_list. When a reference video is included, the unit price is 1.5x the no-reference price (+50%).

Use Cases

• E-commerce & Product Showcase: Upload product images via image_list, reference them in prompts using <<<image_1>>> placeholders to generate product usage scenarios, feature demos, and selling-point videos.
• Short Video & Social Media: Auto multi-shot composition (intelligence) for story-driven content on TikTok, Kuaishou, Xiaohongshu, Bilibili; customize mode for precise creative control.
• Advertising & Marketing: Brand story films, product ads, and campaign videos with optional audio generation and Pro/4K quality.
• Film & VFX Concept Design: Multi-shot composition and reference video style transfer for pre-production concept validation and previz.
• Games & Virtual Worlds: Use element_list preset character library to generate game cutscenes and character demos with consistent visual style.
• First/Last Frame Control: Specify start/end frames via first_frame and end_frame for precise shot transitions and scene changes.

Model Comparison

Kling V3 Omni vs Kling Video O1

Kling V3 Omni vs Kling V3 T2V, Kling V3 I2V

Why Choose Kling V3 Omni?

Choose Kling V3 Omni when video production requires mixed multimodal input (image + video + character), multi-shot narrative composition, reference video style transfer, or precise first/last frame control. It is the only interface in the Kling series covering all these capabilities. A single endpoint supports the full pipeline from text to 4K without switching between different interfaces.

Through iCreat, teams can call V3 Omni via an asynchronous task queue, billed by output video duration. Failed tasks incur no cost. Use Standard (720p) for cost control during testing, then switch to Pro or 4K for premium delivery after confirming the approach.

API

Kling V3 Omni is called through an asynchronous task queue. The workflow is: submit a request to https://api.icreat.ai/v1/task/submit/kwaivgi/kling-v3-omni to obtain a task_id, poll https://api.icreat.ai/v1/task/query-status until the status becomes SUCCEEDED, then call https://api.icreat.ai/v1/task/get-result with the same task_id to retrieve the generated video. Authenticate with your iCreat API Key; request Content-Type is application/json.

To call Kling Video O1 instead, replace the capability code in the endpoint with kwaivgi/kling-video-o1 — both share the same request structure and parameters. Prompts can reference image_list materials via <<<image_1>>> placeholders. For single-shot generation, set multi_shot to false and provide prompt; for intelligent multi-shot, set multi_shot: true and shot_type: intelligence with a prompt; for custom multi-shot, set multi_shot: true and shot_type: customize with 1-6 multi_prompt entries.

Specifications

Top-Level Fields

image_list[] Item

element_list[] Item

video_list[] Item

watermark_info

Model Specifications

Architecture

Kling V3 Omni adopts Kuaishou's Kling-series video generation architecture, unifying text prompts, image materials, video materials, and character materials into a single encoded representation fed into the generation network. The model processes multimodal input tokens within a single context, decomposing long narratives into shot sequences via a multi-shot planning module, then generating and composing shots into a complete video.

Unlike Kling V3 T2V/I2V's single-shot generation, V3 Omni has built-in multi-shot composition capability, completing shot planning and auto-splicing at the model level without requiring developers to call T2V/I2V multiple times and manually edit. Reference videos are processed through a style encoder that extracts color grading, lighting, and cinematography features, injecting them into the generation process for style transfer.

Notes

In complex scenes, multiple small elements may exhibit detail drift between shots — use more detailed prompts or reference videos to constrain the model.

In customize multi-shot mode, each multi_prompt entry should include sufficient cinematography and subject detail to avoid abrupt shot transitions.

When video_list is present in the request, sound can only be set to off — new audio cannot be generated simultaneously.

Non-Chinese/English prompts are untested and may reduce generation quality.

The same prompt produces different variants on each run (by design). For deterministic output, use more detailed prompts or reference videos to constrain the model.

Evaluate based on complete task performance, including visual quality, shot coherence, audio sync, generation latency, and cost. Use Standard for initial prompt and composition testing, then switch to Pro or 4K for delivery after confirming the approach.

FAQ

What's the difference between Kling V3 Omni and Kling Video O1?

Both share the same request structure, parameters, and pricing. On iCreat, they are called via kwaivgi/kling-v3-omni and kwaivgi/kling-video-o1 endpoints respectively. Choose the capability code based on your needs.

What's the difference between Kling V3 Omni and Kling V3 (T2V/I2V)?

Kling V3 Omni is a full-featured interface: supports image/video/character references, multi-shot composition, audio generation, negative prompts, watermark control, and aspect ratio selection. Kling V3 T2V is text-to-video only (prompt + mode + duration), V3 I2V adds an image field on top of T2V — neither supports Omni's advanced features, and max duration is 15 seconds.

When should I use a reference video?

Use refer_type: feature to maintain visual consistency across a content series (color grading, lighting, cinematography style); use refer_type: base to rewrite a scene or replace a subject while preserving the original motion. When a reference video is included, the unit price is 1.5x the no-reference price (+50%).

How many shots does multi-shot mode support?

shot_type: intelligence auto-plans 3-6 shots; shot_type: customize supports 1-6 manual multi_prompt entries.

Should I choose Standard, Pro, or 4K?

Use Standard (std) for testing motion, composition, and prompts at the lowest cost; switch to Pro after confirming the approach; use 4K only for premium deliveries like brand films and TV commercials.

Will I be charged for failed generations?

No. Failed tasks (FAILED) incur no cost and can be safely resubmitted. Regeneration after SUCCEEDED is a new request and will be billed again.