AUTOMATIC1111 hardware, explained.

AUTOMATIC1111, commonly called A1111, is an open-source web interface for Stable Diffusion, named after its original developer. It provides a form-based browser UI with text fields and sliders for the prompt, negative prompt, sampler, steps, CFG scale, seed, and resolution, which made it the most popular and approachable way to run Stable Diffusion for years. A1111 supports txt2img, img2img, inpainting, outpainting, ControlNet, LoRA, embeddings, upscaling, and a large extension ecosystem. It runs SD 1.5, SDXL, and with extensions, Flux. Note that the original A1111 repository's updates have slowed, and many users now run the Forge fork, which keeps the same familiar interface while adding better performance and faster support for new models.

AUTOMATIC1111 is lightweight itself, the hardware demand comes from the Stable Diffusion models you run through it. The key factor is GPU VRAM: 6-8GB runs SD 1.5, 12-16GB runs SDXL comfortably, and 24GB or more is ideal for Flux and heavy extension use. A1111 includes memory-saving flags (--medvram, --lowvram) that let it run on smaller cards at the cost of speed. Pair the GPU with a modern multi-core CPU, 16-32GB system RAM minimum (64GB comfortable), and fast NVMe storage for model checkpoints. NVIDIA GPUs are strongly recommended because A1111 and its extensions target CUDA first, with AMD support being possible but more difficult to set up. Browse AUTOMATIC1111-ready workstations at vrlatech.com/vrla-tech-workstations/generative-ai-workstation.

VRAM is the deciding factor for AUTOMATIC1111. For SD 1.5 and casual SDXL use, NVIDIA RTX 5070 Ti or RTX 5080 (16GB) is good value. For comfortable SDXL, Flux, batch generation, and LoRA training, NVIDIA RTX 5090 32GB is the enthusiast sweet spot, fast generation with room for large models, ControlNet, and the Hires fix upscaling that A1111 users rely on. For professional studios doing high-volume work or running alongside training, NVIDIA RTX PRO 6000 Blackwell 96GB provides the most VRAM available plus ECC memory. NVIDIA is strongly preferred because A1111 and its extension ecosystem target CUDA. If you want the best performance from the same interface, consider running the Forge fork, which is faster than the original A1111 on the same hardware.

VRAM needs scale with the model and resolution. Rough guidance: SD 1.5 at 512x512 runs in 4-6GB, SDXL at 1024x1024 wants 10-12GB comfortably (8GB possible with --medvram), and Flux wants 16-24GB for full quality. A1111's Hires fix upscaling, ControlNet, and multiple LoRAs all add to VRAM use. A1111 includes the --medvram and --lowvram launch flags that trade speed for lower memory, letting a 6-8GB card run SDXL slowly. A practical rule: 8GB is the realistic floor for SDXL with optimizations, 16GB handles SDXL comfortably, and 24GB handles Flux and heavy extension stacks. For the smoothest experience without memory-saving compromises, 16GB or more is the target.

AUTOMATIC1111 and ComfyUI are both interfaces for running Stable Diffusion, but they take opposite approaches. A1111 uses a form-based web UI with tabs, fields, and sliders, you type a prompt, adjust settings, and click Generate. It is approachable and fast to learn, ideal for beginners and straightforward image generation. ComfyUI uses a node graph that you wire together, which is more complex but offers far more control over the pipeline and adopts new models faster. Many people start on A1111 for its simplicity and move to ComfyUI for advanced workflows like video and complex ControlNet stacks. Both run the same models on the same NVIDIA hardware, so a workstation built for one runs the other. The hardware requirements are essentially identical, the choice is about workflow preference.

Forge (Stable Diffusion WebUI Forge) is a fork of AUTOMATIC1111 that keeps the same familiar form-based interface while significantly improving performance and memory efficiency, and adding faster support for newer models like Flux. Because the original A1111 repository's update pace has slowed, many users have migrated to Forge to get a more actively maintained experience without learning a new interface. If you like the A1111 layout but want better speed and current model support, Forge is the natural choice and runs on the same hardware. VRLA Tech workstations can ship with A1111, Forge, ComfyUI, and the Diffusers library all pre-installed, so you can use whichever interface fits each project. The hardware guidance on this page applies equally to A1111 and Forge.

Yes. AUTOMATIC1111 runs SD 1.5 and SDXL natively, and supports Flux through extensions or by using the Forge fork, which has better built-in Flux support. SDXL in A1111 wants 10-12GB VRAM for comfortable 1024x1024 generation, and Flux wants 16-24GB for full quality, with quantized FP8 and GGUF Flux variants running in 8-12GB. The original A1111 was slower to add Flux support than ComfyUI and Forge, which is one reason the Forge fork has become popular among A1111 users who want the newest models. On a workstation with an RTX 5090 32GB, both SDXL and Flux run comfortably in A1111 or Forge with room for ControlNet and upscaling.

AUTOMATIC1111 is GPU-bound, so the CPU does not need to be extreme but should not bottleneck the GPU. A modern AMD Ryzen 9 9950X (16 cores) or Ryzen 7 is plenty for single-GPU setups, handling preprocessing, VAE decoding, and the web server. System RAM of 16GB is the bare minimum, 32GB is comfortable, and 64GB suits heavy use with multiple models and extensions loaded. Fast NVMe storage matters because checkpoints are large (an SDXL model is 6GB, a Flux model is up to 24GB) and switching models loads them from disk, so a fast Gen4 NVMe makes model swapping quick. For most A1111 users, 2TB of fast NVMe is a sensible baseline that holds many models, LoRAs, and ControlNet files.

AUTOMATIC1111 itself focuses on generation rather than training, though it can use LoRAs, embeddings, and hypernetworks at generation time. For training LoRAs, most users run dedicated tools on the same workstation: Kohya_ss is the standard, with OneTrainer and ai-toolkit as alternatives. There is also a training tab and extensions within A1111 for basic embedding and hypernetwork training. LoRA training for SDXL wants 16-24GB VRAM and Flux LoRA training wants 24GB or more. A workstation with an RTX 5090 32GB comfortably trains LoRAs in Kohya_ss and then uses them in A1111 generation, a common creative loop. VRLA Tech workstations ship with A1111, Forge, and the training tools pre-installed so the full workflow works out of the box.

Cloud GPU services for running A1111 cost $0.50-$4 per hour, and several hosted A1111 services exist. For occasional hobby use, cloud is fine. For artists, studios, and businesses generating at volume, a dedicated workstation pays back quickly: unlimited generation with no per-hour fees, full freedom to install any model, extension, or LoRA (hosted services often restrict these), complete privacy for client work, no queue times, and the fast iteration loop that creative work demands. Running A1111 locally also means your models, settings, and outputs are always available and never reset between sessions. For anyone doing Stable Diffusion through A1111 professionally or as a serious hobby, a local workstation is the clear choice. Browse configurations at vrlatech.com.

The best AUTOMATIC1111 workstation in 2026 prioritizes a high-VRAM NVIDIA GPU, since VRAM determines which models and resolutions you can run. For enthusiasts and most creative users, a single NVIDIA RTX 5090 32GB with AMD Ryzen 9 9950X, 64GB RAM, and 2TB NVMe is the sweet spot, fast SDXL and Flux generation, comfortable LoRA training, and headroom for ControlNet and upscaling. For studios doing high-volume commercial work, NVIDIA RTX PRO 6000 Blackwell 96GB paired with Threadripper and 128GB RAM is ideal. For entry-level and SD 1.5 focused use, RTX 5070 Ti or RTX 5080 16GB is a strong value. The same builds run the Forge fork for better performance from the same interface. Browse all configurations at vrlatech.com/vrla-tech-workstations/generative-ai-workstation.

VRLA Tech designs and hand-assembles custom AUTOMATIC1111 and AI image generation workstations in Los Angeles. Browse AI workstation configurations at vrlatech.com/vrla-tech-workstations/generative-ai-workstation. Every system ships with AUTOMATIC1111, the Forge fork, ComfyUI, the Diffusers library, the training tools (Kohya_ss, OneTrainer), CUDA, and xFormers pre-installed and GPU-optimized, a 3-year parts warranty, and lifetime US-based engineer support from engineers who understand creative AI workflows. VRLA Tech is based in Los Angeles and works with creative studios, VFX houses, design teams, and individual artists, alongside enterprise and research customers including Los Alamos National Laboratory, Johns Hopkins University, and George Washington University.