| # PPT Master local configuration generated for this sandbox | |
| # LLM: OpenRouter free model requested by user | |
| OPENROUTER_API_KEY=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 | |
| OPENROUTER_MODEL=tencent/hy3-preview:free | |
| OPENROUTER_BASE_URL=https://openrouter.ai/api/v1 | |
| OPENROUTER_REASONING_EXCLUDE=true | |
| # Image generation: local ComfyUI backend | |
| IMAGE_BACKEND=comfyui | |
| COMFYUI_SERVER=http://127.0.0.1:8188 | |
| COMFYUI_DIR=/app/ComfyUI | |
| COMFYUI_AUTO_START=true | |
| # GPU mode for NVIDIA RTX cards (including RTX 5070 Ti): leave COMFYUI_EXTRA_ARGS empty. | |
| # To force CPU manually, set COMFYUI_EXTRA_ARGS=--cpu. | |
| COMFYUI_EXTRA_ARGS= | |
| # In this CPU sandbox there are no SD checkpoints installed, so placeholder fallback | |
| # keeps the ppt-master workflow executable. On your RTX 5070 Ti machine, set COMFYUI_MODEL | |
| # or COMFYUI_WORKFLOW and COMFYUI_FALLBACK=error for real CUDA rendering. | |
| COMFYUI_FALLBACK=placeholder | |