News

OpenAI Releases gpt-image-1 Model via API for Developer Integration

• By John K. Waters
• April 23, 2025

OpenAI has made its gpt-image-1 image generation model available through its public API, giving software developers direct access to the same system that powers image creation in ChatGPT. The release allows developers to embed high-quality image generation into their applications, workflows, and tools using a simple API call.

The model is natively multimodal, meaning it can generate images from natural language prompts, perform visual edits, restyle images, and render accurate embedded text. It’s optimized for flexibility and visual fidelity, making it suitable for use cases ranging from thumbnail generation and design prototyping to diagramming and ecommerce asset creation.

According to OpenAI, gpt-image-1 supports tasks such as:

• Generating and editing illustrations from text
• Modifying object properties within images
• Applying consistent visual styles
• Inserting readable, stylized text directly into images

The API provides developers with fine-grained control over content filtering through a moderation parameter and includes C2PA metadata for image provenance. OpenAI confirms that it does not use API-generated images or prompts for model training, and all usage is subject to its standard API policies.

Several major platforms have already integrated the model into production:

• Adobe is embedding gpt-image-1 into Firefly and Express, expanding creative options for users in its design suite.
• Figma now supports prompt-based image generation and editing inside Figma Design, enabling rapid visual iteration.
• Airtable has deployed it for enterprise marketing workflows, from concepting to localization at scale.
• Gamma uses the model to generate over 5 million presentation graphics per day, including data visualizations and diagrams.
• HeyGen is applying it for avatar customization and feature editing.
• OpusClip powers one-click YouTube thumbnail creation with prompt-driven personalization.
• Quora has adopted the model as its default image generator to improve prompt alignment and image quality.
• Photoroom leverages gpt-image-1 for staging product photos and generating lifestyle images for ecommerce sellers.

Ongoing pilots with Canva, GoDaddy, HubSpot, Instacart, and Invideo suggest continued expansion into use cases like logo generation, marketing automation, recipe illustration, and AI-powered video editing.

The model is available globally via the OpenAI Images API, with future support planned in the Responses API. Access may require organization-level verification in some cases. Image generation costs are token-based, with output pricing roughly ranging from $0.02 to $0.19 per image, depending on quality settings and output size.

For developers, OpenAI offers documentation, a playground for testing prompts, and integration guidance to get started with gpt-image-1. The API supports real-time generation and customization, offering flexible integration options for both frontend and backend applications.