Z-Image Studio

This server wraps the Z-Image-Turbo diffusers model and exposes it via MCP tools for local AI agent workflows. You get image generation with seed control, model listing with hardware-aware precision recommendations, and generation history browsing. It's optimized for local hardware acceleration across NVIDIA CUDA, AMD ROCm on Linux, Apple Silicon MPS, and Intel XPU. Supports multiple LoRAs per generation and automatic dimension adjustment. The MCP implementation runs over stdio, SSE, or Streamable HTTP transports, returning identical structured content across all three. Reach for this when you want Claude or another agent to generate images locally without hitting external APIs, especially if you're already running the Z-Image CLI or web UI and want to expose those capabilities programmatically.