GPT-Image-2.5 in Production: Choosing Between Flare and Sunburst

Description and use

<h2> The short version </h2> <p>GPT-Image-2.5 is split into two models:</p> <ul> <li> <strong>Flare</strong> favors low latency and routine generation.</li> <li> <strong>Sunburst</strong> favors editing precision, reference preservation, and higher-fidelity final output.</li> </ul> <p>Both accept text and image inputs, support the quality levels <code>auto</code>, <code>low</code>, <code>medium</code>, <code>high</code>, <code>xhigh</code>, and <code>max</code>, and work through the Images API for generation and editing.</p> <p>For an existing OpenAI-compatible integration, the migration is sm

Problem

<h2> The short version </h2> <p>GPT-Image-2.5 is split into two models:</p> <ul> <li> <strong>Flare</strong> favors low latency and routine generation.</li> <li> <strong>Sunburst</strong> favors editing precision, reference preservation, and higher-fidelity final output.</li> </ul> <p>Both accept text and image inputs, support the quality levels <code>auto</code>, <code>low</code>, <code>medium</code>, <code>high</code>, <code>xhigh</code>, and <code>max</code>, and work through the Images API for generation and editing.</p> <p>For an existing OpenAI-compatible integration, the migration is sm

Outcome

<h2> The short version </h2> <p>GPT-Image-2.5 is split into two models:</p> <ul> <li> <strong>Flare</strong> favors low latency and routine generation.</li> <li> <strong>Sunburst</strong> favors editing precision, reference preservation, and higher-fidelity final output.</li> </ul> <p>Both accept text and image inputs, support the quality levels <code>auto</code>, <code>low</code>, <code>medium</code>, <code>high</code>, <code>xhigh</code>, and <code>max</code>, and work through the Images API for generation and editing.</p> <p>For an existing OpenAI-compatible integration, the migration is sm

Evidence

GPT-Image-2.5 in Production: Choosing Between Flare and Sunburst

Source