Supported Models

Many different model types exists in the generative space. Invoke support many of the most popular and effective types out of the box. 

Base Models

Invoke supports the state of the art base models, and will continue to expand support as new models are released. Invoke includes a set popular and effective models default, while Premier and Enterprise plan users have the option to add additional models for use with Invoke.

Currently supported base model types are:

• Stable Diffusion XL (SDXL)
• Stable Diffusion 1.5 (SD1.5)
• FLUX.1 - find additional guidance on Flux model support here.

 If you're unfamiliar with models, learn more here.

LoRAs

Invoke supports the use of LoRAs and LyCORIS for supported Stable Diffusion base models (SD1.5, SDXL). To see a list of currently supported Flux LoRAs, please visit the Flux Supported Models section here. Invoke provides commonly used LoRAs by default, with Premier and Enterprise plan accounts able to add new LoRAs to be used with Invoke.

If you're unfamiliar with LoRAs, learn more here.

ControlNet, IP Adapter, and T2I Adapter Models

You can upload custom ControlNet, IP Adapter, and T2I Adapter Models that are trained on similar/common architectures & with standard inference pipelines that match publicly available models. You can also upload Flux Tools like Depth and Canny through the "import starter models" button in the Model Manager tab (requires attestation of commercial license).

If you have a model that is not working, please feel free to reach out to our team at [email protected] to discuss your use-case.

Adding Models and LoRAs

If you are interested adding new models or LoRAs to use with Invoke, you can learn more here.