OpenAI: GPT Image 1 Mini

GPT Image 1 Mini is an image generation model from OpenAI. It is positioned as a cost-efficient variant of GPT Image 1, aimed at producing high-quality images with reduced latency compared to the full model. On api14 it appears under the image category and is identified by the model id openai/gpt-image-1-mini.

The model accepts text and image inputs and returns images. Its context window is 400,000 tokens, with a maximum output of 360,000 tokens. Because the output modality is image rather than text, the model is built for generation tasks rather than chat or text completion. Image inputs mean you can supply reference material alongside your prompt, which the model can use to guide what it produces.

That combination of text and image input with image output supports a range of practical work. You can generate images from written descriptions, produce variations on an existing image by passing it as a reference, or steer a generation with several references at once. The aspect ratio, quality and background controls let you match output to a specific layout or use case, such as a square social asset, a portrait crop or a transparent background for compositing. Because generation is the only output, the model is not intended for tasks like captioning, classification or text extraction from images.

You call it through api14's OpenAI-compatible API using the model id openai/gpt-image-1-mini. The supported request parameters are max_tokens, response_format, seed and structured_outputs. Alongside these, the model exposes image-specific controls: aspect_ratio accepts 1:1, 3:2, 2:3 or auto; quality accepts auto, low, medium or high; background accepts auto, transparent or opaque; n sets how many images to return, from 1 to 10; input_references sets how many reference images to pass, from 0 to 16; and output_compression runs from 0 to 100. A seed lets you reproduce a generation, and response_format controls how the result is returned.

Keep the parameter set in mind when planning an integration. The model outputs images only, so if you need text back from an image, a vision or multimodal text model is the better fit. The 400,000 token context window is generous for prompts and references, but the output is an image, not prose. If your workload centers on generating images from text or references and you want the cost-efficient variant of GPT Image 1, this model is the appropriate choice.