Open-source app for AI image generation with a canvas and node workflows
Invoke is a free, open-source app for making and editing images with AI models. It runs on your own computer as a local web server that you use in the browser. It has a layer-based canvas for drawing, painting and editing (inpainting and outpainting), node-based workflows you can build and share, a gallery with boards, and a model manager for checkpoints, LoRAs and ControlNets.
It runs local models such as SD 1.5, SDXL, FLUX.1, FLUX.2 Klein, Z-Image, Qwen Image and Krea 2. It can also call third-party APIs (OpenAI, Google Gemini, BytePlus Seedream, Alibaba Cloud), billed to your own account. Video with Wan 2.2 is experimental.
The software is free under the Apache 2.0 licence. The Invoke Launcher installs it on Windows 10+, macOS 14+ (Apple Silicon) and Linux; Docker and manual installs also work.
The official hosted platform has been shut down: the docs say the founding team joined Adobe. The project is now maintained by the community, led by Lincoln Stein and Vic (Blessedcoolant).
Creatives, professionals and enthusiasts who want to generate and edit images with AI on their own hardware and keep their data private. The docs say it is built by creatives, for creatives.
A layered canvas for drawing, inpainting and outpainting, next to node workflows.
A layered canvas, node workflows and a model manager, with new releases in 2026 (6.14.2 in September 2026).
Free and runs on your own computer, with a layered canvas for editing; Midjourney is a paid cloud subscription from $10 a month.
Open-source node-based app for AI images, video, audio and 3D
Makes video, audio and 3D as well as images, with 661 ready-made templates. Comfy Cloud runs it in the browser; InvokeAI's hosted platform has shut down.
AI service that makes images and short videos from text prompts
A hosted service: no install, no models to download, and any computer works.
Discussion
Sign in to join the discussion