Core capability
Official capabilities announced by Kuaishou
Kuaishou’s announcement says Video 3.0 supports text-to-video, image-to-video, reference-to-video, and in-video editing in one multimodal workflow. It describes image and video references for consistency, multilingual audio, multi-shot instructions, and video generation up to 15 seconds. The same announcement assigns 2K and 4K output to Image 3.0 and Image 3.0 Omni; it does not establish those resolutions for Video 3.0.
Artly availability
Kling 3.0 is available in Artly’s custom generator. The production allowlist includes Kling 3.0 standard, Pro, and 4K text-to-video and image-to-video IDs. No curated exact-version template is currently attached to this article; generator availability and curated-template availability are different states.
Artly test and alternative evidence
No curated Kling 3.0 article test gallery is attached or verified. Kuaishou’s release visuals are official demonstrations, not Artly test results. If you generate with Kling 3.0, evaluate the result under the selected Kling 3.0 mode and its own prompt and input.
Evidence
Prompt guidance and use cases
For a Kling 3.0 custom-generator brief, number the beats and give each one a subject action, camera direction, and transition. Keep reference images purposeful: identify which one defines the subject and which one defines the setting. This is editorial prompt guidance, not evidence that every mode will produce the same result.
Compare related video models
Related pages include the Wan AI Video Generator guide, the Seedance AI Video Generator guide, the Wan 2.7 release explainer, and the Seedance 2.5 release explainer.