How is Flare different from Sunburst?

Speed against precision. Flare is tuned for volume: fast, high quality, the everyday choice. Sunburst takes longer and gives you tighter control across edits, for hero shots and campaign creative.

Is Flare better than GPT Image 2?

Yes, on quality, speed, reference fidelity and editing precision, and it costs the same 3 to 9 credits. GPT Image 2 stays available as the previous generation, and older API ids resolve to it.

How much does Flare cost on APImage?

3 to 9 credits per image, the same band as GPT Image 2, and it is included on every plan down to the free tier. Quality xhigh costs twice that and max three times. The exact figure for your settings is shown before you generate.

Does it still render text well?

Yes. Flare inherits the text rendering that made GPT Image 2 the pick for anything with words in the frame, and improves on the real world information embedded in images generally.

Do new models cost extra on APImage?

Never. Every model is in your plan, priced in the same credits out of the same pool. New models simply appear, and Ultra gets them first.

Twice as fastOpenAI

GPT Image 2.5 Flare.
Better, at half the wait.

OpenAI's September 2026 release, and the new default for most work. Higher quality than GPT Image 2 at 50 percent lower latency, for the same 3 to 9 credits, and reference handling good enough to carry a face across a whole campaign.

Generate with Flare See pricing

3 to 9 credits per image. Included on every plan, including the free tier. Quality xhigh costs 2x and max 3x.

OpenAI

GPT Image 2.5 Flare

What is GPT Image 2.5 Flare?

GPT Image 2.5 Flare is one of two API models OpenAI shipped alongside ChatGPT Images 2.5 in September 2026, and it is the one built for volume. It produces higher quality images than GPT Image 2 at roughly half the latency, and it is markedly better at working from reference photos: a subject moved into a new setting, style or composition stays recognisably that subject, with lighting and texture that read as real. Localized editing is the other change, so a fix to one corner of an image no longer quietly redraws the rest of it. Flare is the default choice for most applications on APImage, with Sunburst reserved for the shots that need extra precision.

One subscription, every model

One pool of credits

No per model surcharge

API and MCP access included

Specifications

Provider: OpenAI
Released: September 2026
Positioning: Default for most applications
API model ID: gpt-image-2.5-flare, aliased from gpt-image-2.5
Latency: Around 50 percent lower than GPT Image 2
Quality: Higher than GPT Image 2 across the board
Reference photos: Stronger subject fidelity across settings and styles
Editing: Localized, with multi turn consistency
Text rendering: Inherits GPT Image 2's strength
Plans: Every plan, including free
Cost: 3 to 9 credits per image, 2x at xhigh, 3x at max

Why this model

What GPT Image 2.5 Flare does that the others do not.

Half the wait

Around 50 percent lower latency than GPT Image 2, at higher quality. When you are generating a hundred variants, that is the difference between a session and an afternoon.

References that hold

Move a familiar subject into a new setting, style or composition and it stays recognisable. Distinctive features carry through instead of being averaged away.

Edits that stay local

Change one region and the rest of the frame survives untouched, across several turns of edits. The old failure mode was fixing a hand and losing the background.

Light and texture that read

Materials and lighting behave more naturally, which is what separates an image you can ship from one that is merely correct.

What people make with it

Where it earns its credits in real work.

High volume social

Twenty variants of a post, generated in the time the previous model took for ten.

Rapid prototyping

Explore a direction at speed, and only slow down for the frame that is going to be the hero.

Consistent brand faces

One reference set, many scenes, and the same person in all of them.

Iterative retouching

Fix it, look at it, fix the next thing. Localized editing makes turn three as clean as turn one.

Catalog imagery

A product in twelve contexts, from one set of references, without twelve photoshoots.

How it works

1Open the image studio and pick GPT Image 2.5 Flare.
2Upload your reference photos. This is the model that rewards good references.
3Describe the scene, and quote any text that has to appear in the frame.
4Generate, then refine with localized edits rather than starting over.

Create your account and try it

Or use one of these

GPT Image 2.5 Flare is not the right answer to everything. Your plan includes all of them, so switching model costs nothing.

GPT Image 2.5 Sunburst

The shot is the hero and you want tighter control across edits, at the cost of speed.

See GPT Image 2.5 Sunburst

GPT Image 2

You want the cheapest OpenAI option and the extra speed is not worth the credits.

See GPT Image 2

Flux 2 Max

No text in the frame, and you want maximum photographic detail.

What it costs

3 to 9 credits per image

Included on every plan, including the free tier. Quality xhigh costs 2x and max 3x.

No separate subscription

GPT Image 2.5 Flare is on the same plan as every other model on APImage, priced in the same credits out of the same pool. Run the same prompt through three models and keep the one that looks right.