Draw Things

Draw Things is an AI-assisted image and video generation application designed around private, on-device creative workflows. It allows users to generate and edit visual content directly on Apple devices, including iPhone, iPad, and Mac, without requiring a constant internet connection for local generation. The platform is particularly focused on making advanced generative AI models practical on consumer hardware while giving users control over their models, prompts, images, and creative workflow.

One of Draw Things’ defining characteristics is its emphasis on local AI processing. The free edition allows users to download supported models and generate images using the computing resources of their own device. This makes the platform attractive to creators who prefer keeping their creative work on-device rather than sending every generation to a remote server. Draw Things also provides cloud computing options for users who need additional processing power.

The application began as an effort to bring sophisticated image generation to Apple’s mobile ecosystem. Draw Things was already running full image-generation models directly on mobile hardware by 2023, and its engineering work has increasingly focused on optimizing inference for Apple Silicon and Apple’s Neural Engine.

Draw Things supports a broad range of modern generative models. Its release history shows continuing support for models and technologies including Stable Diffusion, SDXL, FLUX, Qwen Image, Ideogram, Krea, Wan, LTX, ERNIE Image, Z Image and other open-weight models. The platform is therefore more than a basic text-to-image application: it acts as a flexible environment for experimenting with different AI models and creative techniques.

For image generation, users can enter natural-language prompts and generate images locally. The application also supports image editing and transformation workflows, allowing creators to work from existing images rather than starting from text alone. Its capabilities include image-to-image workflows, inpainting, outpainting, ControlNet-related workflows, upscaling, image enhancement and other model-specific controls.

Another major feature is Infinite Canvas. Rather than restricting users to a fixed image boundary, Draw Things provides an expandable creative workspace where users can extend compositions and generate additional content around an existing image. This can be useful for concept development, illustration, visual storytelling, social-media artwork, product concepts and other design projects.

Draw Things has also expanded beyond still images. The current platform supports image-to-video and multiple video-generation models. Its release history includes support for Wan 2.1 and Wan 2.2, Hunyuan Video and LTX-2.3, allowing users to experiment with AI-generated motion and video directly within the application.

Personalization is another important part of the platform. Draw Things supports LoRA training, allowing users to fine-tune models for specific characters, visual concepts or artistic styles. Importantly, LoRA training can be performed locally. Draw Things introduced on-device fine-tuning for Stable Diffusion models on iPhone, iPad and Mac in 2023, making model customization possible without relying entirely on dedicated cloud GPUs.

The application is also designed with Apple hardware optimization in mind. Its engineering team has developed custom inference technologies, including Metal FlashAttention, to improve generation performance on Apple Silicon. Later releases added further optimization for Apple Neural Engine hardware and newer M-series chips. On compatible M5 hardware, Draw Things introduced Lightning Draft for faster interactive generation with supported models.

Draw Things is useful for digital artists, illustrators, designers, photographers, marketers, researchers and AI enthusiasts who want detailed control over generative media. It can be used for concept art, character development, image editing, visual experimentation, marketing graphics, product concepts, story development and creative research.

The platform also offers a cloud-based option for users whose devices cannot comfortably run larger models. The Community edition provides cloud compute, while Draw Things+ provides increased compute limits, custom LoRA support, additional privacy features and other premium cloud capabilities. Enterprise customers can deploy Draw Things across organizational computers or servers and can integrate local inference or cloud compute into applications.

For developers, Draw Things provides scripting capabilities and an API-oriented ecosystem. Its documentation describes JavaScript scripts that can extend the application, while its newer MediaGenerationKit provides a Swift package for integrating open-weight image and video models into applications with local or cloud inference.

Privacy is a central part of the product’s positioning. Local generation can take place directly on the user’s device, while cloud features are optional. Draw Things’ privacy documentation explains how account information, cloud-generated content and subscriptions are handled when users choose cloud services.

Overall, Draw Things combines local AI generation, advanced model support, image editing, video generation, model customization and optional cloud computing in one Apple-focused creative application. Its strongest distinction is the ability to run increasingly capable generative models locally on iPhone, iPad and Mac while still providing cloud and developer options when additional computing resources are needed.

Draw Things offers a Free Edition with local image generation, model downloads, Infinite Canvas and local LoRA training. A Community tier provides cloud compute and Lab Hours, while Draw Things+ costs $8.99/month and adds higher cloud-compute limits, custom LoRA uploads, increased compute capacity and Privacy Pass. Enterprise pricing is available through a custom arrangement for organizations.

Review: Draw Things is particularly suitable for users who want private, local AI image generation rather than relying entirely on cloud services. Its extensive model support, Apple hardware optimization, editing tools, LoRA training and growing video capabilities make it a flexible creative platform. The main limitation is its strong dependence on Apple hardware for its primary experience, while larger models can also require substantial device memory and processing resources.