Stable Diffusion WebUI Codex

FastAPI + Vue 3 WebUI for running multiple diffusion engines with explicit contracts (no silent fallbacks).

✨ What you get

• 🧠 Multi-engine: SD15, SDXL, FLUX.1, Chroma, Z-Image, and WAN 2.2 (video).
• 📡 Streaming jobs: request → task_id → SSE events → results.
• 🧾 Explicit contracts: missing/invalid inputs fail loud (no silent guessing).
• 🧰 Utility views: PNG Info, XYZ plot, GGUF tools, Workflows snapshots, Settings.

🚀 Quick start

Prereqs:

• Git
• Internet access for first install (downloads uv, CPython, Node.js, and wheels)
• Optional: ffmpeg + ffprobe on PATH (video export / vid2vid)

Windows

install-webui.bat
run-webui.bat

Linux / WSL

bash install-webui.sh
./run-webui.sh

When the WebUI starts, it prints the URLs:

[webui] API: http://localhost:7850
[webui]  UI: http://localhost:7860

Open the UI URL in your browser. Stop with Ctrl+C.

More details (CUDA/ROCm selection, troubleshooting): see INSTALL.md.

📦 Models

Official Hugging Face repo: https://huggingface.co/sangoi-exe/sd-webui-codex

Available weights on the Hub (generated via hf download --dry-run on 2026-01-29):

• ⚡ FLUX.1
  • flux/FLUX.1-dev-Q5_K_M-Codex.gguf
• 🖼️ Z-Image
  • zimage/Z-Image-Turbo-Q5_K_M-Codex.gguf
  • zimage/Z-Image-Turbo-Q8_0-Codex.gguf
• 🎞️ WAN 2.2
  • wan22/wan22_i2v_14b_HN_lx2v_4step-Q4_K_M-Codex.gguf
  • wan22/wan22_i2v_14b_LN_lx2v_4step-Q4_K_M-Codex.gguf
  • wan22-loras/wan22_i2v_14b_HN_lx2v_4step_lora_rank64_1022.safetensors
  • wan22-loras/wan22_i2v_14b_LN_lx2v_4step_lora_rank64_1022.safetensors
  • wan22-loras/wan22_t2v_14b_HN_lx2v_4step_lora_rank64_1017.safetensors
  • wan22-loras/wan22_t2v_14b_LN_lx2v_4step_lora_rank64_1017.safetensors

Coming soon (supported by the WebUI, but not published on the Hub yet):

• ⏳ Chroma (Flux.1 Chroma packs)
• ⏳ Z-Image Base variant packs (Turbo is available today)
• ⏳ WAN 2.2 5B packs (GGUF)
• ⏳ WAN 2.2 Animate packs (GGUF)
• ⏳ More WAN 2.2 packs/variants

Refresh the list:

hf download --dry-run --quiet sangoi-exe/sd-webui-codex --include '*.gguf' --include '*.safetensors'

By default, models live under models/:

• models/sd15/, models/sdxl/, models/flux/, models/zimage/, models/wan22/
• plus *-vae, *-tenc, *-loras

Model hub notes and folder layout: see README_HF_MODELS.md.

If you customize model roots, edit apps/paths.json.

📄 License (noncommercial)

• Code license: PolyForm Noncommercial License 1.0.0 (LICENSE).
• Required notice must be preserved: NOTICE.
• Commercial use is not permitted: COMMERCIAL.md.
• Trademarks/branding: TRADEMARKS.md.

🙏 Credits

• Semantic baseline for “parity”: Hugging Face Diffusers.
• Inspiration: AUTOMATIC1111 + Forge (Codex backend/UI are reimplemented; no copy/paste ports).