MindCraft Studio is a local-first creative powerhouse. To begin your journey, you first need to "fuel" your studio with high-performance generation engines.

Finding Your Engines

The Model Library section in the Settings Modal is your command center. From here, you can discover, install, and manage your AI model library.

Featured Models

• Flux.2 [Klein]: A premium 4-step distilled model designed for extreme speed without sacrificing quality. It is particularly excellent for high-fidelity image editing and rapid iteration.
• Z-Image Turbo: A versatile 8-step engine optimized for photorealism and bilingual (English & Chinese) text rendering. It features robust instruction adherence for complex compositions.

To install a model, simply click the Install button on its card.

Settings
APPLICATION
GeneralStorage
MODEL LIBRARY
Base ModelsLoRAUpscalerBackground RemoverAssistant
Base Models
Foundational models for generating images
Import
Starter Models
Z-Image Turbo (4-bit)
(~6GB)
FLUX.2 Klein 4B (4-bit)
(~4.6GB)
Install

Importing Custom Models

MindCraft Studio is built on Apple's MLX framework, deeply optimized for the Neural Engine on Apple Silicon. You can import custom MLX-converted models (Supports Z-Image Turbo, Flux.1, Flux.2 [Klein] and SDXL models)

You can browse a collection of supported architectures on the Models Explorer.

Local Import

If you have compatible MLX model files already downloaded, click Import from Local. Select the root folder of the model; MindCraft will automatically detect the architecture, verify the weights, and register it in your library.

Remote Import

You can also use the Import from Remote button to import compatible models from the Hugging Face / ModelScope directly. Simply enter the repository ID (e.g., themindstudio/flux-klein-4b-mlx-4bit) and click import. MindCraft handles the download and optimization in the background.