ChatGPT Images 2.5 Pricing: Plans, API Rates, and the Real Cost Drivers

The short answer: ChatGPT Images 2.5 is included across ChatGPT plans, with different usage levels, while the API bills separately for text and image tokens. In the captured U.S.-dollar pricing view, Free is $0 per month, Go is $8, Plus is $20, and Pro starts at $100. For the API, GPT-Image-2.5 Flare and Sunburst currently share the same published unit rates—but that does not guarantee the same final cost for every generated image.

OpenAI announced ChatGPT Images 2.5 on September 8, 2026, positioning the release around sharper detail, faster generation, and more precise editing.

ChatGPT Images 2.5 Pricing at a Glance

Plan prices and features can vary by country, currency, tax treatment, and future product changes. The figures above reproduce the official view supplied for this article and should be rechecked before publication or purchase.

OpenAI’s ChatGPT pricing page, captured in a French-language locale with U.S.-dollar monthly prices, shows how image access scales from Free through Pro.

GPT Image 2.5 API Pricing

OpenAI publishes the same token rates for both GPT-Image-2.5 variants. Image tokens cover image inputs and generated image outputs; text tokens cover the prompt and other text passed to the model.

Important: these are prices per one million tokens, not fixed per-image fees. The amount charged for an individual request depends on the tokens actually consumed.

The official model catalog confirms matching text-input prices for Flare and Sunburst and support for both image generation and image editing endpoints.

Why There Is No Single GPT Image 2.5 Cost per Image

A headline such as “one image costs X” is only reliable when the model, dimensions, quality, input assets, and request path are all fixed. The API can charge different totals because of:

Output quality. The API supports auto, low, medium, high, xhigh, and max. Higher-quality outputs generally require more image output tokens.

Output dimensions. Square, portrait, landscape, and high-resolution custom sizes do not necessarily consume the same number of tokens.

Reference images and edits. Image inputs add image-input tokens; repeated editing turns can add further input and output usage.

Streaming previews. Each partial image requested during streaming adds 100 image-output tokens.

The API path. When image generation is called as a tool in the Responses API, the mainline text model may also generate billable tokens.

Retries and rejected drafts. A campaign budget should count all generations, not only the final files that get published.

Do Flare and Sunburst Cost the Same?

At the rate-card level, yes. OpenAI currently lists the same input and output token prices for GPT-Image-2.5 Flare and Sunburst. At the job level, not necessarily. Flare is optimized for speed, while Sunburst is optimized for maximum quality and tighter editing control. If one model consumes a different number of output tokens, uses a different quality setting, or needs more retries for your specific brief, the final invoice can differ even when the unit rates match.

ChatGPT Subscription vs. API Billing

How to Estimate a Real Project Budget

A practical budget should be built from a representative request, not from a generic cost-per-image number.

1.  Choose Flare or Sunburst and lock the intended quality and dimensions.

2.  Define the typical number of reference images and whether the job is generation, editing, or a multi-turn workflow.

3.  Run a small representative batch when the model is available in your production environment, then record the actual token usage.

4.  Calculate cost per usable output, including retries, alternate versions, and partial images—not just cost per API call.

5.  Add the text-model cost when the Responses API or another orchestration layer is part of the workflow.

For third-party creative platforms: credits are platform-specific packages, not a direct translation of OpenAI’s token rate. If GPT Image 2.5 appears in a platform such as Astorie, check the live model selector and quoted credit cost before publishing a conversion table.

Frequently Asked Questions

Is ChatGPT Images 2.5 free?

It is available through the Free plan with limited and slower image generation. Paid plans provide higher or more advanced access.

Is ChatGPT Images 2.5 included in Plus?

Yes. OpenAI lists more advanced and more precise image generation among Plus features.

Does ChatGPT charge per image?

The consumer ChatGPT product is sold by subscription tier, not through a published per-image checkout price. Usage limits still apply.

Is Flare cheaper than Sunburst?

Their published API unit rates are currently identical. Flare can still be more economical in a workflow if its speed and output are sufficient with fewer costly iterations, but that must be measured.

What does the API bill?

Text input, cached text input, image input, cached image input, and generated image output tokens. Additional text-model usage can apply in the Responses API.

Which plan should a creator choose?

Free is suitable for occasional exploration; Go or Plus suits more frequent creation; Pro is aimed at heavier use. The right choice depends on your actual volume and whether you need API automation.

Bottom Line

ChatGPT Images 2.5 does not have one universal price. ChatGPT users pay by plan, while API users pay by token consumption. Flare and Sunburst share the same published unit rates, so the meaningful comparison is cost per acceptable result under your own dimensions, quality setting, reference inputs, and retry rate.

Official Sources

OpenAI launch  |  ChatGPT pricing  |  Image generation guide  |  API pricing