GPT Image 2.5 Flare
For everyday images, quick concepts and frequent iterations. OpenAI positions Flare as the faster choice.
Create a new image. Refine the details that matter.
GPT Image 2.5 is not yet available to generate on AIVG. Explore Flare and Sunburst below, or create with our available image models.
Explore available image modelsFor everyday images, quick concepts and frequent iterations. OpenAI positions Flare as the faster choice.
For detailed product imagery and focused edits where preserving the surrounding composition matters most.
Both accept text and reference images and produce images. These are image models; they do not generate video or audio.

Describe the product to retain, then specify the setting and lighting to change.
Provide exact text, visual hierarchy and the intended format. Check spelling in the result.
Name the element to change and the details to preserve, then review each revision.
Writing examples to help you start; these are not claims of tested results.
Keep the ceramic cup from my reference image. Place it on a pale stone shelf, lit by soft morning light. Preserve its shape, glaze and handle.
Create a poster for “NIGHT GARDEN”. Make the event title the largest text. Use a botanical illustration, a dark background and a clear space for the date.
Change only the jacket to deep blue. Preserve the face, pose, background and lighting.
Yes. OpenAI announced ChatGPT Images 2.5 on September 8, 2026, with the Flare and Sunburst API models.
Not yet. This page introduces the model. You can currently create with the available models listed in the image workspace.
API charges depend on tokens, image settings and reference inputs. AIVG pricing for this model has not been published. GPT Image 2 prices should not be used to estimate a 2.5 generation.
No. Sketch and comment-based editing are ChatGPT features. Their announcement does not mean they are available in the AIVG interface.
Reviewed September 9, 2026