LongCat Video
Features
- Generates video with a 13.6 billion parameter diffusion Transformer
- Keeps characters and appearance consistent across a video with Block Sparse Attention
- Two-stage generation: fast 480p draft, then refinement to 720p or 1080p at 30fps
- Continues clips seamlessly from their last frames
- Handles fluid motion and camera moves like pans, dollies and orbits
- Runs on local GPUs or through cloud APIs
- Starts from a single character image and extends it into a sequence
Pricing Details
💵 Pricing & Plans (February 2026 Status): Plan Price (Approx.) Key Features Open Source $0 (Self-Hosted) Free for local use (requires Nvidia A100/H100 or 4090 with optimizations). Creator Cloud ~$30 / mo Access to high-speed cloud rendering; 1080p output; no watermark. Studio / API Usage-based High-priority rendering; custom fine-tuning on your own characters/assets. 🎁 How to Get Started: For developers, the model weights are available on Hugging Face. For creators, you can access the cloud interface at LongCat.video. Start by uploading a single image of a character and use the "Extend" feature to build a 60-second cinematic sequence around them.