Unlimited image and audio generation on your hardware.
ALICE runs Stable Diffusion and audio generation models locally. No subscription, no cloud upload, no cost per image or track. Generate whatever you want - your prompts and media never leave your machine.
Free and open source. macOS (MPS), Linux (CUDA, ROCm, Vulkan), and Steam Deck. Deploy standalone or via Docker.
Everything you need to create.
Generation studio
Text-to-image and image-to-image with full parameter control. LoRA models, negative prompts, multiple schedulers, seed control, and resolution presets.
Private gallery
Masonry grid gallery with per-image privacy controls. Optional time-limited public sharing. Regenerate from any image, view metadata, batch manage your collection.
Model marketplace
Browse, search, and download models from CivitAI and HuggingFace directly in the UI. SDXL, SD 1.5, FLUX, GGUF, and custom fine-tunes all supported.
Real-time dashboard
GPU memory, system metrics, generation history, and quick actions. Monitor everything from a single view.
Multi-GPU support
NVIDIA CUDA, AMD Vulkan/ROCm, Apple MPS, and CPU fallback. Dual backend architecture: PyTorch for full diffusers pipeline support, and stable-diffusion.cpp with Vulkan for universal GPU acceleration without vendor drivers.
Admin and authentication
User management, API key generation, role-based access. Deploy for yourself or share with your team.
Steam Deck ready
Dedicated SteamOS installer with user-level systemd service. Run Stable Diffusion on your Steam Deck's AMD GPU via ROCm or Vulkan - no desktop Linux required.
Docker deployment
Pre-built containers for CPU, CUDA, and ROCm on GitHub Container Registry. One command to run: docker compose --profile cuda up -d.
A prompting guide built right in.
ALICE includes model-specific prompting guidance so you get better results from the start. Learn what works for SD 1.5, SDXL, and FLUX without leaving the app.
Adjust steps, guidance scale, denoising strength, samplers, and schedulers. Use LoRA models for style transfer. Full Stable Diffusion capabilities without compromise.
Your images, your rules.
Every image is private by default. Share individual images with optional time-limited public links that expire automatically. NSFW filtering is built in and configurable.
All generation happens on your hardware. No cloud services, no data collection, no usage tracking. Run on macOS or Linux - your GPU, your data, your creations.
Deploy anywhere.
Run standalone, as a systemd or launchd service, or in Docker. Self-update system keeps ALICE current automatically. Built-in SteamOS/Steam Deck installer for AMD GPUs.
OpenAI-compatible REST API for integration with SAM, scripts, or any tool. Generate images and audio from conversations, automate workflows, or build on top of the API.
Part of the ecosystem
ALICE + SAM Chat-Driven Generation
SAM connects to ALICE for image and audio generation. Ask SAM to "create an image" or "generate a song" and ALICE handles it via OpenAI-compatible API. Seamless, private, local.
ALICE + CLIO Automated Deployment
CLIO orchestrates ALICE deployments across fleets. Remote GPU provisioning, model management, benchmarking, Docker/ROCm configuration - all automated from your terminal.
OpenAI-Compatible API Universal Integration
Any client that speaks the OpenAI image or audio API can use ALICE. Plug into existing workflows, custom scripts, or third-party tools without modification.
Get started with ALICE.
Download for macOS or Linux. Generate images locally.