I was checking the pricing for the new GPT-Image-2.5 models and noticed that both Flare and Sunburst appear to have exactly the same token pricing.
From the documentation, I understand that Flare is mainly optimized for faster everyday generation, while Sunburst is aimed at higher precision and more detailed creative or editing work.
This made me wonder why they are exposed as two separate models instead of having one GPT-Image-2.5 model with something like a parameter for speed vs precision.
For example, something like:
mode: "fast"
or
mode: "precision"
Or perhaps this could be handled through the existing quality parameter.
Since the token pricing is the same, having two model names makes the choice slightly confusing from a developer point of view.
Is there a deeper technical difference between Flare and Sunburst that makes separate model IDs necessary?
Are there situations where Sunburst consumes more output tokens or ends up costing more per generated image even though the per-token rate is identical?
I am mainly trying to understand when a developer should explicitly choose one over the other instead of simply using Flare for speed and Sunburst for accuracy.
