InvokeAI
invoke.ai
- Free plan
- Windows
- macOS
- Linux
- API
- Self-hosted
What InvokeAI is
InvokeAI is a free, open-source creative engine for AI image generation that runs locally on the user's hardware. Its layer-based canvas supports drawing, painting, sketching, and image editing, while a visual node editor helps build repeatable generation workflows. Listed models include Flux, SDXL, and SD 1.5, with a Model Manager for checkpoints, LoRAs, Textual Inversions, and ControlNets. Other tools include ControlNet guidance, batch generation, prompt wildcards, and high-resolution upscaling. InvokeAI offers API access and PNG output. External integrations include Google Gemini, OpenAI, BytePlus Seedream, and Alibaba Cloud DashScope, with usage billed by the provider; those external providers do not currently support inpainting. The project says local use keeps data, prompts, and creations under the user's control. The software is licensed under Apache 2.0. Hosted GPU options are linked through AI Badgr, RunPod, and Railway. The former Invoke.ai hosted platform has shut down, while the open-source project continues under community stewardship.
Who it suits
InvokeAI suits people who want to generate and edit images with a locally run, open-source tool. It may also suit teams or individuals building repeatable workflows, provided they account for provider billing when using external models.
Good to know, and what to watch
- Free and open-source under Apache 2.0
- Layer-based canvas for image editing
- Visual node editor for repeatable workflows
- Supports batch generation and high-resolution upscaling
- Runs locally on the user's hardware
- Former Invoke.ai hosted platform has shut down
- External model usage is billed by providers
- External providers do not support inpainting
- Software has not undergone rigorous security audit or intrusion testing
Plans
Every plan and priceInvokeAI
Free plan
The verdict
InvokeAI offers a broad local image-generation and editing toolkit, plus API access and hosted GPU links. Consider the provider billing and stated security-audit limitation when deciding how to deploy it.
What the maker publishes
- API access
- Yes source
- Canvas
- Its layer-based canvas supports drawing, painting, sketching, and editing images. source
- External model limit
- Inpainting is not currently supported by any external provider. source
- External providers
- External model integrations include Google Gemini, OpenAI, BytePlus Seedream, and Alibaba Cloud DashScope; usage is billed by the provider. source
- Free plan
- Yes source
- Generation tools
- Features listed include ControlNet guidance, batch generation, prompt wildcards, and high-resolution upscaling. source
- Hosted options
- The download page links to hosted GPU options on AI Badgr, RunPod, and Railway. source
- Hosted service status
- The former Invoke.ai hosted platform has shut down, while the open-source project continues under community stewardship. source
- Image editing
- Yes source
- License
- Invoke is free and open-source under the Apache 2.0 license. source
- Models
- The site lists support for Flux, SDXL, SD 1.5, and other models, with a Model Manager for checkpoints, LoRAs, Textual Inversions, and ControlNets. source
- Output formats
- PNG source
- Platforms
- Download options are listed for Windows, macOS, and Linux; the page specifies Windows 10 or later, Apple Silicon for macOS, and an NVIDIA or AMD GPU for Windows and Linux. source
- Privacy
- Invoke runs locally on the user’s hardware, and its site says data, prompts, and creations remain under the user’s control. source
- Purpose
- Invoke is a creative engine for AI-powered image generation. source
- Reference images
- Yes source
- Security
- The administrator guide says the software has not undergone a rigorous security audit or intrusion testing. source
- Support
- The project points users to its FAQ, Discord community, documentation, and GitHub Issues for help. source
- Text-to-image
- Yes source
- Workflows
- A visual node editor lets users build complex, repeatable image-generation workflows. source
Checked against the maker's pages on 9 Oct 2026.