comfyui-mcp/packs/z-image-base/pack.yaml

Local-first, agent-native control plane for ComfyUI — MCP server + autonomous sidebar agent that drives your live graph in natural language on ANY LLM: Claude/ChatGPT/Gemini on your subscription (no API key), free local models via Ollama (fully offline),

name: z-image-base
display_name: Z-Image Base
family: z-image
kind: txt2img
description: >-
  Z-Image Base — the 6B text-to-image base model (BASE, not Turbo). Trainable and
  finetuning-friendly, with more diverse styles than the distilled Turbo variant.
  V2 GGUF-quantized workflow runs in <8GB VRAM and pairs the base model with a
  Z-Image-Fun ControlNet Union, the Detailer Daemon, and an 8-step distill LoRA.
  Z-Image Base is the finetuning-friendly counterpart to z-image-turbo.
vram: "<8GB"
workflow: workflow.json
skill: z-image-txt2img
launch_args: []
sources:
  mirror: https://huggingface.co/Aitrepreneur/FLX
  model: https://huggingface.co/Aitrepreneur/FLX
notes:
  - >-
    Run the generated installer from your ComfyUI root (the folder containing
    custom_nodes/ and models/). The installer offers a Q5_K_S / Q6_K / Q8_0 GGUF
    tier menu; the shipped workflow pins Q8_0.
  - >-
    Z-Image Base is the trainable, more-diverse counterpart to z-image-turbo —
    the workflow ships both loaders, with the base model as the default.
  - >-
    ControlNet preprocessor weights (depth_anything_v2_vitl, dw-ll_ucoco,
    yolox_l) are not bundled; comfyui_controlnet_aux downloads them on first use.
  - >-
    If you hit "Torch not compiled with CUDA enabled", reinstall a CUDA build of
    torch into the embedded/venv python (the portable build can ship CPU-only torch).
  - >-
    Pascal/Maxwell GPUs (GTX 10xx / 9xx) need the cu126 torch build — newer cu12x
    wheels drop those compute capabilities and fall back to CPU.
post_install:
  - >-
    Restart ComfyUI (or use ComfyUI-Manager -> Install missing custom nodes for
    any node Python deps).
  - Load workflow.json.