A quicker, cheaper member of the gpt-image family. It can inpaint using a mask but offers neither input_fidelity nor dedicated preservation of faces.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains gpt-image-1-mini in context, with comparison tables and the common traps.
Terms in this definition
- Image edit mask
A PNG matching the input image's dimensions in which fully transparent pixels (alpha 0) mark the sole region the model is allowed to alter.
- Dedicated
Running Azure Functions on an App Service plan, which removes the execution time limit and offers VNet integration on Basic and higher tiers.
Related terms
- gpt-image-1
OpenAI's text-to-image model series, which took over from DALL-E in Azure. The first gpt-image-1 preview is retiring on 23 October 2026, so newer models such as gpt-image-2 or gpt-image-1-mini are used for new projects.
- input_fidelity
A parameter on GPT-image edits, low by default or high, that keeps the features and style of input images, like a product's identity or faces, intact while leaving unrelated areas alone. gpt-image-1-mini does not support it.