Stable Diffusion is an open source latent diffusion model for text-to-image generation. It runs locally on consumer GPUs, supports inpainting, outpainting, and image-to-image workflows. Community fine-tunes and LoRAs expand styles endlessly. Licensed openly, it powers countless creative tools, research experiments, and production pipelines worldwide today and beyond.
Open Source DiffusionDALL-E PyTorch is an open replication of OpenAI's DALL-E, built entirely in PyTorch. It generates images from text prompts using a discrete VAE and transformer. Researchers use it to study multimodal generation, experiment with architectures, and reproduce results without proprietary APIs or restricted access today.
DALL E ReplicaFlux is an open model family for high-quality text-to-image generation. It uses flow matching and rectified flows for sharp, detailed outputs. Variants range from schnell for speed to pro for quality. Open weights enable local inference, fine-tuning, and integration into creative pipelines everywhere.
Flux Open ModelFooocus is a simple, open source user interface for Stable Diffusion. It removes complex settings, offering prompt enhancement, style presets, and quality defaults. Designed for beginners and professionals alike, it runs offline, keeps workflows fast, and produces strong images with minimal configuration required.
Fooocus Simple UIComfyUI is a node-based interface for Stable Diffusion and other models. Users build workflows by connecting nodes, enabling advanced control, batching, and automation. It is highly extensible, supports custom nodes, and suits power users seeking reproducible, flexible image generation pipelines today.
ComfyUI Node BasedSDXL is a high-resolution open diffusion model producing detailed, photorealistic images. It improves composition, lighting, and text rendering over earlier versions. With larger parameters and better conditioning, it excels at 1024px outputs and supports fine-tuning for specialized domains and creative work.
SDXL High ResKandinsky is an open multimodal model for text-to-image generation. It combines a text encoder with a diffusion decoder, producing diverse, creative images. Released for research and community use, it supports experimentation, localization, and integration into open creative tooling and pipelines.
Kandinsky Open ModelDeepFloyd IF is an open cascaded diffusion model for text-to-image generation. It uses multiple stages to refine resolution and detail, with strong prompt adherence. Designed for research, it supports high-quality outputs and experimentation with large-scale generative architectures and methods.
DeepFloyd IF Model32 questions, each with clarification and a suggested next step. Keep a separate draft for each generator. This page records a plan; it does not install software or generate images.
Unanswered questions remain visible. Review open questions before acting on the plan; recorded answers do not certify an installation.
Answers stay in this browser unless you export them. Do not enter passwords, API keys or other credentials.