GPT Image 2.5

Flare or Sunburst?

It is speed versus precision, not two quality tiers. OpenAI positions Flare as the default for most work and Sunburst as the option for edit-heavy, production-ready creative, which takes longer to generate. On Klox they accept the same aspect ratios and resolution tiers and cost the same credits, so running one prompt through both is a cheap way to decide. The figures below are read from the live model configuration.

What this generation is built for

• Reference photos into new scenes

  Carrying a person, a pet or a product from a reference photo into a new setting, style or composition, with distinctive features and natural lighting surviving the move.

• Subject consistency

  Swapping a product, a background or a line of copy while the subject, the composition and everything around it stays put. This is the improvement OpenAI leads with, and it is what makes an edit usable instead of a fresh roll of the dice.

• Edits that build on each other

  Feed a result back in as the reference for the next node and earlier changes hold instead of degrading. Each step is its own node on the canvas, so you can branch from a good state rather than lose it.

• Complex briefs and layouts

  Detailed art direction — hierarchy, composition, a defined visual style — is more likely to survive as the instructions get more specific, which is what on-brand asset series and UI concepts need.

Frequently asked questions

What is the difference between GPT Image 2.5 Flare and Sunburst?

Speed versus precision. OpenAI describes Flare as the default for most applications, generating faster than GPT Image 2, and Sunburst as an extra level of precision for detailed creative work, with longer generation times. On Klox both take the same options and cost the same credits, so the cheapest way to choose is to run the same prompt through each.

Is this the same as ChatGPT Images 2.5?

Same generation, different names. ChatGPT Images 2.5 is what the model is called inside ChatGPT; Flare and Sunburst are the two models OpenAI released for the API, and those are the ones Klox calls. ChatGPT-only features such as Sketch and templates are not part of the API models.

Do I need an OpenAI API key?

No. Klox calls the model for you and bills in credits, so there is no key to obtain, no separate account and no per-request billing to manage.

How many credits does a GPT Image 2.5 generation cost?

It depends on the resolution tier you pick, and the table above shows the current figure for each. The generator also shows the exact cost before you run it.

Can I use my own photos as references?

Yes, and it is what this generation is best at. Attach reference images to the prompt and the model works from them for subject, style or composition; the table above shows the current limit per generation.

What happens to the images I generate here?

Each generation lands on a Klox canvas as a node, so you can regenerate it, adjust the prompt, hand it to another model, or use it as the first frame of a video shot.