Jaaz is an open-source, privacy-first multimodal creative agent for generating images and videos from text prompts or chat messages. It supports local execution via ComfyUI alongside optional cloud-based generation models.
Project overview
Jaaz is an open-source, privacy-first multimodal creative agent for generating images and videos from text prompts or chat messages. It supports local execution via ComfyUI alongside optional cloud-based generation models.
Project type
AI Agent · Image & Vision · Video
Use cases
Video Creation
Deployment
Refer to project documentation
License
License pending
Best for
Creators and general users who prioritize local-first, offline-capable AI media generation.
Users seeking a hybrid setup to combine private local execution with optional cloud-based media generation.
Key capabilities
Turn a single prompt into complete images or videos in seconds with auto-optimized prompts and multi-turn refinement.
Local-first architecture ensuring no data leaves the device, with no tracking.
Limitations and risks
Manual installation on Linux requires Python >=3.12 and manual command-line execution.
The repository license is NOASSERTION, creating potential ambiguity for open-source reuse.
GPU requirements are not documented by the project.
Getting started
Download the application directly from https://jaaz.app or GitHub releases for immediate use on Windows or macOS. The setup difficulty is rated as easy due to the downloadable desktop binary.
Linux users must install from source, which requires Python version 3.12 or higher and manual command-line execution.
Evidence and sources
GitHub project description: The world's first open-source multimodal creative assistant This is a substitute for Canva and Manus that prioritizes privacy and is usable locally.
README: The world's first open-source multimodal canvas creative agent
README: This is a substitute for Canva and Manus that prioritizes privacy and is usable locally.
README: Enterprise Cloud “Full” Edition</b> — Private/on-prem deployment & commercial licensing (Docker image or full source). Includes all jaaz.app online features.
README: One-Prompt Image & Video Generation Turn one prompt into complete images or videos in seconds. -Supports GPT-4o, Midjourney, VEO3, Kling,veo3,seedance etc.