OpenAI: GPT Image 1

GPT Image 1 is an image generation and editing model from OpenAI, listed on api14 under the model id openai/gpt-image-1. It sits in the image category, and its provider describes it as generating and editing images through a dedicated Images API. If you already work with OpenAI models, this is the same vendor, exposed here through api14's OpenAI-compatible surface.

The model takes text and image inputs and returns images. Its context window is 400,000 tokens, with a maximum output of 360,000. The image parameters give you control over the shape and quality of what comes back: aspect_ratio accepts 1:1, 3:2, 2:3 or auto; quality accepts auto, low, medium or high; background accepts auto, transparent or opaque. You can request between 1 and 10 images per call with n, supply between 0 and 16 input references for edits, and set output_compression anywhere from 0 to 100. The provider highlights accurate text rendering, transparent backgrounds, and support for up to 16 reference images when editing.

Those capabilities point to a fairly specific set of jobs. Accurate text rendering makes it a reasonable fit for images that need legible words inside them, such as banners, mockups or product shots with labels. Transparent backgrounds suit assets that will be composited elsewhere, like icons, logos or layered design elements. The reference image input is what makes editing practical: you can pass existing images and ask for changes rather than generating from scratch, which is useful for iterating on a design or restyling an asset while keeping its structure. Generating several variations in one call with n helps when you want options to choose from rather than a single result.

To call it through api14, use the model id openai/gpt-image-1 against the OpenAI-compatible endpoint, so existing OpenAI client libraries and request shapes should carry over with minimal changes. The supported parameters are max_tokens, response_format, seed, and structured_outputs, alongside the image parameters described above. Use seed when you need reproducible output across runs, and response_format to control how the result is returned. Because the model outputs images rather than text, treat it as a generation endpoint rather than a chat completion one, even though the API surface is compatible.

One practical consideration is that the output modality is images only. If your task is text generation, classification, or anything that returns prose or structured text, a text model will fit better; structured_outputs here applies within an image generation request rather than turning the model into a text assistant. The 400,000 token context and 360,000 maximum output are generous, but the useful constraint is usually the 16 reference image ceiling for edits and the 10 image ceiling per call. Plan around those numbers when batching work.