News

OpenAI Brings Image Generation Model to Its API, Enabling Broader Developer Access

• By John K. Waters
• 04/23/2025

OpenAI has made its latest image generation model, gpt-image-1, available through its API, enabling developers and businesses to integrate image generation into their applications and platforms.

The model, which underpins image generation in ChatGPT, is natively multimodal, which means it can generate high-quality images from text prompts, follow detailed instructions, and incorporate both real-world knowledge and visual fidelity. It supports a range of visual tasks, including illustration, editing, restyling, and rendering text within images. Since its debut in ChatGPT last month, the feature has seen widespread use, the company says, with users generating more than 700 million images in the first week.

The model uses the same safety guardrails as the image generation feature in ChatGPT, the company says, including C2PA metadata for provenance and content filters. Developers can adjust moderation sensitivity via an API parameter. OpenAI has said that customer data from API usage is not used to train its models, and that all inputs and outputs are governed by standard API use policies.

The API release allows for image generation to be directly embedded in products across industries such as creative tools, e-commerce, education, gaming, and enterprise software. OpenAI noted in a blog post that early adopters include well-known platforms such as Adobe, Figma, Airtable, and Wix, among others.

Adobe is incorporating gpt-image-1 into its Firefly and Express tools, the company says, providing creative professionals with additional visual style options for brainstorming and design. Figma has begun rolling out features in Figma Design that allow users to generate and edit images directly within the design environment, supporting visual experimentation and faster iteration. And Airtable is applying the model in enterprise workflows, such as campaign asset generation, remixing, and localization, streamlining creative production at scale.

Other platforms leveraging gpt-image-1 to enhance their visual capabilities include Gamma, which uses the model to generate more than five million images daily for diagrams, visualizations, and design variations; HeyGen, which uses the model to improve avatar customization; OpusClip, which enables YouTube creators to produce thumbnails with a single click; Quora, which has has adopted it as its default image generator; Photoroom, which integrates the model into tools for staging and lifestyle imagery tailored to e-commerce sellers.

Other companies currently experimenting with integrations include Canva, GoDaddy, HubSpot, Instacart, and Invideo, each exploring use cases in areas such as logo creation, marketing collateral, recipe illustration, and video editing.

The model is now available globally via OpenAI’s Images API, with support coming soon to the Responses API. Organizations may be required to verify access before use.