# GPT Image 2

GPT Image 2 is OpenAI's image generation and editing model from April 2026, the generation before GPT Image 2.5. It is built for precise edits, legible text in images and flexible sizes. In Sloosh it takes up to 16 references and renders at 1K, 2K or 4K.

Web page: https://sloosh.ai/models/gpt-image-2

[Use GPT Image 2](https://sloosh.ai/studio/image?image-model=gpt-image-2)

[View templates](https://sloosh.ai/models/gpt-image-2/templates.md)

## Key facts in Sloosh

|   |   |
| --- | --- |
| Made by | OpenAI |
| Makes | image |
| Inputs | text, up to 16 reference images |
| Resolution | 1K, 2K or 4K (default 1K) |
| Aspect ratios | Auto, 1:1, 3:2, 2:3, 16:9, 9:16, 4:5, 5:4 (default Auto). At 2K or 4K, Auto renders square (Sloosh docs). |
| Quality | low, medium, high (default medium) |
| Prompt limit | 32,000 characters |

## Best for

- Designed layouts with several blocks of text, such as recipe cards, menus and posters. OpenAI lists text rendering as a GPT Image 2 strength ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Refining one image over several edits, passing each output back in ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Edits from photo inputs: OpenAI says image inputs are always processed at high fidelity on GPT Image 2 ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Brand boards and social kits: the Brand Board Space ([sloosh.ai/spaces/brand-board](https://sloosh.ai/spaces/brand-board)) and Social Profile Kit Space ([sloosh.ai/spaces/social-profile-kit](https://sloosh.ai/spaces/social-profile-kit)) fit this job.

## How to use GPT Image 2 in Sloosh

1. Open [Image Studio with GPT Image 2 selected](https://sloosh.ai/studio/image?image-model=gpt-image-2), or pick GPT Image 2 in the model chip.
2. Add up to 16 references if needed.
3. Write the prompt in the order the image is laid out, with each block's text in quotes.
4. Set aspect ratio, resolution and quality (use high for small text).
5. Check the cost on the Generate button, then generate.

## GPT Image 2 prompting tips

- Use labeled sections (scene, subject, details, constraints) for complex requests ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Lay the prompt out top to bottom like the image, each block's text in quotes; short lines render more reliably than long paragraphs ([Sloosh docs](https://sloosh.ai/docs/gpt-image-2.md)).
- Quote exact text, describe its position and typography, and spell brand names letter by letter if needed ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Use medium or high quality for small text or dense information ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).
- Treat camera specs as appearance cues, not physics ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).

## GPT Image 2 example prompts

**Prompt 1**

```text
A clean, modern recipe card for an iced matcha latte, for a cafe's Instagram. Top: a photo of the drink in a tall glass with ice. Below it, the title "ICED MATCHA LATTE" in a bold rounded font. Then three numbered steps in a simple sans-serif: "1. Whisk matcha with hot water" "2. Pour over ice and milk" "3. Stir and serve". Cream background, green accents.
```

**Prompt 2**

```text
A one-page cafe menu, portrait. Header "THE CORNER CUP" in a bold serif. Two columns: "COFFEE" and "TEA", four items each with a price on the right. Thin divider lines, warm off-white paper, dark brown text. No other text.
```

**Prompt 3**

```text
Image 1 is a logo. Make a flat lay of a business card, an envelope and a letterhead on a light grey desk, each printed with the logo from image 1. Keep the logo's shape and colours exactly as in image 1. Soft overhead light.
```

## How it compares in Sloosh

| Model | Resolution | Quality levels | References | Prompt |
| --- | --- | --- | --- | --- |
| GPT Image 2 | 1K, 2K, 4K | low, medium, high | 16 | 32,000 |
| [GPT Image 2.5 Flare](https://sloosh.ai/models/gpt-image-2.5-flare) | 1K, 2K, 4K | low to max (5) | 16 | 32,000 |
| [GPT Image 2.5 Sunburst](https://sloosh.ai/models/gpt-image-2.5-sunburst) | 1K, 2K, 4K | low to max (5) | 16 | 32,000 |
| [Ideogram 4.5](https://sloosh.ai/models/ideogram-4.5) | 1K, 2K | auto, low, medium, high | 5 | 5,000 |

OpenAI's guide: if GPT Image 2 meets your needs, test Flare for latency; if it falls short on a complex task, start with Sunburst ([OpenAI](https://developers.openai.com/api/docs/guides/image-prompting)).

## Where else it runs

- OpenAI API, model gpt-image-2 (OpenAI model page, accessed 2026-10-30).

## Frequently asked questions

### What is GPT Image 2?

OpenAI's image generation and editing model (released 21 April 2026), described by OpenAI as built for fast, high-quality generation and editing. It powers ChatGPT Images 2.0.

### How do I use GPT Image 2?

In Sloosh, open Image Studio, pick GPT Image 2, add up to 16 references, set ratio, resolution and quality, and generate. No API key needed.

### Is GPT Image 2 free?

OpenAI's API charges per image. In Sloosh the exact price shows on the Generate button before you run.

### GPT Image 2 vs GPT Image 2.5?

In Sloosh both take 16 references and up to 4K. GPT Image 2.5 Flare and Sunburst add xhigh and max quality. OpenAI calls Flare its small model, optimized for speed with quality comparable to GPT Image 2, and Sunburst its base model, optimized for quality above GPT Image 2. See the [Sunburst](https://sloosh.ai/models/gpt-image-2.5-sunburst) and [Flare](https://sloosh.ai/models/gpt-image-2.5-flare) pages.

### What is a good GPT Image 2 prompt?

State the result and where it will be used, then subject, setting and light, camera, exact text in quotes, and what must not change.

### Can GPT Image 2 render text?

Yes, OpenAI lists text rendering as a capability. Quote every word and use high quality for small text.

## Related

- Models: [GPT Image 2.5 Flare](https://sloosh.ai/models/gpt-image-2.5-flare), [GPT Image 2.5 Sunburst](https://sloosh.ai/models/gpt-image-2.5-sunburst), [Ideogram 4.5](https://sloosh.ai/models/ideogram-4.5), [Nano Banana Pro](https://sloosh.ai/models/gemini-3-pro-image-preview)
- Spaces: [Brand board](https://sloosh.ai/spaces/brand-board), [Social profile kit](https://sloosh.ai/spaces/social-profile-kit)
- Docs: [GPT Image 2](https://sloosh.ai/docs/gpt-image-2.md), [Prompting GPT Image](https://sloosh.ai/docs/prompting-gpt-image.md)
- Answers: [Which AI image model renders text in images most accurately?](https://geo.sloosh.ai/posts/which-ai-image-model-renders-text-in-images-most-accurately), [What is the best AI image model in 2026?](https://geo.sloosh.ai/posts/what-is-the-best-ai-image-model-in-2026)

## Sources

- [OpenAI: GPT Image 2 model page](https://developers.openai.com/api/docs/models/gpt-image-2)
- [OpenAI: Image prompting guide](https://developers.openai.com/api/docs/guides/image-prompting)
- [Sloosh docs: GPT Image 2](https://sloosh.ai/docs/gpt-image-2)
- [Sloosh pricing](https://sloosh.ai/pricing)