OpenAI has released its latest image generation technology to developers via API. This tool, called ‘gpt-image-1’, is already available to most ChatGPT users.
Now, third-party platforms can integrate it into their apps and services.
The image generator quickly gained attention online in March 2025 as users shared stunning results across social media.
Styles ranged from animated Ghibli-like scenes to realistic “AI action figures.” In just one week, over 130 million users created more than 700 million images.
The rollout attracted millions of new users. However, it also placed pressure on OpenAI’s systems.
Despite the demand, the company is moving forward. Developers now have full access through OpenAI’s API.
What GPT-Image-1 Can Do
GPT-image-1 is OpenAI’s newest image model, and it is natively multimodal, meaning it can process both text and images.
It, however, does more than generate pictures; it understands instructions, adapts to different styles, and applies real-world context.
Key features include:
- Creating art in various styles (e.g., anime, photorealistic, cartoon)
- Following specific prompts (e.g., “sunset over a mountain village”)
- Generating readable and stylized text within images
- Supporting batch creation of multiple images
Major Brands Already Adopting the Tool
Several well-known companies have begun using GPT-image-1. These firms are exploring how the model can enhance their platforms.
| Company | Description |
| Figma | Adds image creation and editing inside the design suite |
| Instacart | Tests AI-generated visuals for recipes and shopping lists |
| Canva, Wix, Adobe | Investigate creative design applications |
| GoDaddy | Considers use in brand and website image generation |
| Airtable | Explores visual support for databases and workflows |
How Developers Can Use the API
The API allows for precise control over the image creation process. Developers can choose the image quality to balance speed and output resolution and create multiple images in one request.
They can also select moderation settings for content control
There are two moderation levels:
- Auto: Applies standard content filtering
- Low: Loosens restrictions slightly while remaining policy-compliant
Every image includes a watermark using C2PA metadata.
This tag allows platforms to identify images as AI-generated.
The Cost
Pricing is based on tokens. Tokens are the units of data the model reads or writes.
| Token Type | Price per 1M Tokens | Approx. Image Cost |
| Text input | $5 | Minimal cost |
| Image input | $10 | ~7 cents for medium quality |
| Image output | $40 | ~2 to 19 cents per image |
Safety and Responsible Use
OpenAI has emphasized safety in this release. GPT-image-1 includes safeguards to prevent harmful or inappropriate content.
Developers can adjust the content filter settings. However, both levels enforce core policy restrictions. The model will not generate content that is violent, explicit, or misleading.
C2PA metadata ensures transparency, therefore, platforms can trace images back to their AI origin.
