gpt-image

The gpt-image skill generates and edits images using OpenAI's GPT-Image-2 model through the inference.sh CLI (belt). It gives users a single model that handles text-to-image generation, image editing with reference images, and mask-based inpainting, solving the need to produce product mockups, marketing visuals, concept art, and photo manipulations from the command line. After belt login, all work is done with belt app run openai/gpt-image-2 and a JSON --input payload.

Capabilities documented include text-to-image, image editing using reference images passed in an images array, mask-based inpainting via a mask image, batch generation of up to 10 images at once (n parameter), multiple output formats (PNG, JPEG, WebP), and flexible resolution at any size in 32px increments from 256 to 4096. Parameters include prompt (required), images, mask, n (1-10), quality (low, medium, or high), width and height (256-4096, multiples of 32), output_format (png, jpeg, or webp), and output_compression (0-100 for jpeg/webp). Pricing is approximately $0.006 per low-quality image, $0.024 medium, and $0.21 high, with larger resolutions costing more; full pricing is available via belt app get openai/gpt-image-2.

The skill targets designers, marketers, and developers who want scripted image generation and editing. It requires the inference.sh CLI (belt), installable via npx skills add belt-sh/cli, and restricts its allowed tool to Bash(belt *). Related skills point to the full inference.sh platform skill, general image-generation models, FLUX models, and Pruna P-Image for faster, more economical generation.