Open-source web studio for generating AI images and videos through 600+ hosted models from one interface.
Open-source local AI engine that runs LLMs, vision, voice, image, and video models behind OpenAI-compatible APIs on CPU or GPU.
Native macOS and Apple Silicon inference server exposing OpenAI, Anthropic and Ollama APIs, with a Swift app, written in Zig.
C/C++ inference engine for diffusion image and video models such as SD, Flux, Wan, and Qwen-Image, built on ggml.