# GPT Image 2.5

Make images from a text prompt or one reference image with GPT Image 2.5 by OpenAI, at 37 credits per image. See its settings, limits and exact prices.

GPT Image 2.5 is an OpenAI image model listed for realistic images and typography. It takes a text prompt or one reference image. It costs 37 credits per image in both modes, well under GPT Image 2 (98 for text, 113 with a reference).

## At a glance

- **Maker:** OpenAI
- **Type:** Image model
- **Modes:** Text to image and image to image
- **Aspect ratios:** 16:9, 4:3, 1:1, 3:4, 9:16. The Studio starts on 4:3.
- **Outputs per run:** 1 to 4, set with **Count**
- **Reference image:** One reference image (JPEG, PNG or WEBP, up to 20 MB). A run with a reference image makes 1 image.
- **Resolution:** No resolution choice

## What it costs

These are credits per image, as of 2026-10-10. The Studio shows the live quote for your settings before you generate, and the total it charges is that quote multiplied by **Count**.

| Setting | Credits per image |
| --- | --- |
| Any aspect ratio | 37 |

The price is the same at every aspect ratio.

## Use it in Studio

1. **Open Create Image** In Studio, open **Create Image**. Sign in when the Studio asks, because every generation needs an account.
2. **Choose GPT Image 2.5** Open the model picker, which groups models by maker, and choose GPT Image 2.5.
3. **Write your prompt** Describe the image. To start from a picture, attach one reference image (JPEG, PNG or WEBP, up to 20 MB).
4. **Set the output** Choose an aspect ratio and **Count** from 1 to 4.
5. **Check the price and generate** The Studio shows the credit price for your settings. It is the same price the server charges. Results appear as cards and are saved to your **Library**.

## Limits

- You need to be signed in to generate. GPT Image 2.5 is available on every plan, including Free; credits are the limit.
- One reference image only, and a run with a reference image makes 1 image, not 1 to 4.
- If a generation fails, the credits are refunded. The Studio does not run a different model in its place.

## Common questions

**How is GPT Image 2.5 different from GPT Image 2.5 Sunburst?**

Both are OpenAI models that take a text prompt or one reference image, and both cost 37 credits per image. The picker tags GPT Image 2.5 as Realistic and Typography, and Sunburst as Realistic and HD.

**Can I use my own image with GPT Image 2.5?**

Yes. Attach one reference image (JPEG, PNG or WEBP, up to 20 MB) before you generate. A run with a reference image makes one image.

**Similar models**

- [GPT Image 2.5 Sunburst](https://www.uova.com/help/models/image/gpt-image-2-5-sunburst): Realistic images, fine detail. Default credits: 37.
- [GPT Image 2](https://www.uova.com/help/models/image/gpt-image-2): Text rendering, photorealism. Default credits: 98 text, 113 with reference.

Source: https://www.uova.com/help/models/image/gpt-image-2-5