Kling O1

Kling O1 is Kuaishou's unified multimodal video model, released December 1, 2025. It's the world’s first AI model combining text prompts, reference images, and video inputs to edit existing footage. Swap, add, or remove elements without frame-by-frame VFX work. The model is built on a Multimodal Visual Language (MVL) framework that interprets all inputs together for consistent edits across frames.

You can access Kling O1 on VEED’s AI playground. With a subscription and AI credits, upload videos, describe edits, and refine in VEED's built-in editor. For full capabilities, including text-to-video and start/end frame control, access Kling O1 directly at klingai.com.

Kling O1 supports various video editing tasks through multimodal prompts:

• Swap outfits or characters while preserving original motion
• Add products into footage for marketing and e-commerce content
• Build scenes by combining multiple reference images with consistent elements
• Remove distracting objects, passersby, or unwanted elements from shots

Kling O1 specs on VEED's AI playground:

• Input video length: 3-10 seconds
• Minimum resolution: 720p
• Output length: 10 seconds
• Reference images: Up to 4 images
• Editing modes: Swap items, Add items, Remove items

On VEED, Kling O1 currently costs 60 credits per second—you'll see the exact credit cost when you select the model. A VEED subscription with AI credits gives you access to Kling O1 alongside other AI video models, plus more AI tools for post-generation work. On Kuaishou's platform (klingai.com), Video O1 is available on paid plans. Check their website for current pricing and offers.

The key differentiator: Kling O1 is the first model to unify text, images, and video inputs for editing in a single workflow. In Kuaishou's internal benchmarks, Kling O1 outperformed Veo 3.1 on image reference tasks and Runway Aleph on video transformation tasks. These are Kuaishou's self-reported results, so we recommend testing for yourself. Other models like Veo 3.1 and Seedance offer multimodal and editing capabilities, but typically through separate modes rather than one unified interface.