OpenAI Brings Image Generation Model to API via gpt-image-1

OpenAI has announced the release of its multimodal image generation model, gpt-image-1, to the API, allowing developers and businesses to integrate image creation capabilities directly into their tools and platforms. The model, previously introduced in ChatGPT, supports image generation from text prompts, detailed editing, and accurate rendering of styles and text.

“When we introduced image generation in ChatGPT last month, it quickly became one of our most popular features,” OpenAI said, citing over 700 million images created by 130 million users in the first week.

The gpt-image-1 model can be used across multiple domains such as e-commerce, design, education, enterprise software, and gaming. It supports functions including transforming sketches into polished visuals, editing existing images, and producing marketing content at scale.

Adobe, Airtable, Figma, Gamma, HeyGen, OpusClip, Quora, Wix, Photoroom, and Playground are among the companies already leveraging the model. 

Adobe, for example, is embedding it in its Firefly and Express apps to allow users to explore different visual styles. Figma users can now edit images within the platform, adjusting elements like objects and backgrounds from text instructions.

Airtable is using the model to scale creative workflows. “Enterprise marketing and creative teams [can] easily manage asset workflows at scale—from generating campaign concepts, to remixing assets, to localising media for international markets,” the company stated.

OpusClip enables YouTube creators to generate thumbnails in one click, while Quora is replacing its existing image model with gpt-image-1 to improve generation quality. Photoroom is using it to create tools like Product Beautifier and Virtual Model for e-commerce visuals.

New use cases are being tested by companies like Canva, GoDaddy, HubSpot, Instacart, and invideo AI. GoDaddy, for instance, is experimenting with logo generation and editable social media assets, while Instacart is exploring its application for recipe visuals and shopping list imagery.

The gpt-image-1 API includes safety features such as harmful image prevention and embedded C2PA metadata. Developers can also adjust content moderation levels using the moderation parameter.

Pricing is based on token usage. Text input costs $5 per million tokens, image input $10 per million tokens, and image output $40 per million tokens. “In practice, this translates to roughly $0.02, $0.07, and $0.19 per generated image for low, medium, and high-quality square images,” OpenAI noted.

The model is now available globally via the Images API. Some organisations may need to verify access before use. Developers can explore capabilities in the API playground and consult the image generation guide to get started.