zhao123xiao

Generate and edit images with Dreamfield's OpenAI-compatible gpt-image-2 image API. Use when the user asks Codex to create, test, or save an image from a text description, make wallpaper or illustration assets, perform image-to-image generation from reference images or clipboard images, use masks/inpainting, verify Dreamfield image generation, or troubleshoot image-generation responses such as empty data, b64_json, URL output, quota errors, or content policy errors.

End-to-end OpenVINO standard operating procedures for model conversion, quantization/compression, deployment, smoke testing, benchmarking, device selection, and troubleshooting. Use when the user asks to deploy a model with OpenVINO, convert or quantize a Hugging Face/ONNX/TensorFlow/PyTorch model to OpenVINO IR, serve models with OpenVINO Model Server or OpenAI-compatible endpoints, test local OpenVINO artifacts, or benchmark CPU/GPU/NPU/AUTO performance.