GPT Image 1.5 in the Responses API: choosing an integration surface

The API update

On December 19, 2025, OpenAI added GPT Image 1.5 and chatgpt-image-latest to the Responses API image-generation tool.

Choose around the workflow

A direct image request fits a form that already knows its image prompt and desired operation. An orchestrated workflow may need to assemble instructions through several steps before producing an image. Start by drawing those steps and identifying where image generation belongs.

Do not change APIs just because a new integration surface exists. A working direct image pipeline may already give your application the control and observability it needs.

Check support at every layer

Provider documentation describes the provider API. Your application must also check which endpoints and request formats its selected service exposes. Support for native Images calls does not automatically establish support for every Responses tool combination.

For XMH.NET, consult the console documentation and available models for the supported native image request. Confirm any additional orchestration requirement before implementing it.

Keep image output handling explicit

Store the generated asset, its source brief, and the request identifier together. Define what happens when an orchestration step succeeds but the image step fails. A clear boundary helps users retry the image operation without accidentally repeating the whole business action.