Models
Video Models
Compare video models on LazyKiwi and run any of them in one workbench.
Sora 2
OpenAI video model for cinematic clips with synchronized audio.
Hailuo 2.3 Fast
MiniMax image-to-video model: upload one frame, get 6 to 60 seconds at 768p or 1080p.
MiniMax H3
MiniMax video model with 2K output and the longest prompt field.
Kling v3 Omni
Kuaishou video model with text, image and start-end frame modes up to 60 seconds.
Wan 2.7 I2V
Alibaba image-to-video and start-end frame model, 2 to 60 seconds at 720p or 1080p.
Wan 2.2
Alibaba MoE image-to-video for short, stable motion from a still.
Seedance 2.0
ByteDance video generation for smooth, social-ready clips.
Seedance 2.5
Newest Seed video model — up to 30s and 21:9 through 9:16.
LTX-2.3
Lightricks open DiT video model for text-to-video and image-to-video.
HunyuanVideo 1.5
Tencent’s compact 8.3B open video model for coherent T2V and I2V.
HappyHorse 1.0
Text and image to video in five aspect ratios, 3 to 60 seconds at up to 1080p.
Grok Imagine Video
xAI video model with the widest ratio set in the workbench, from 21:9 to 4:5.
Veo 3.1 Lite
Google Veo 3.1 Lite for 16:9 and 9:16 clips at 720p or 1080p, 4 to 60 seconds.