Developers Can Now Access OpenAI's Image Models via API

OpenAI has finally made its highly sought-after image generation capabilities available to developers through its API. The company announced today the release of gpt-image-1, the native multimodal model that powers ChatGPT's image creation tool. It is now accessible globally via the Images API, with plans to integrate it into the Responses API soon.

Key Highlights

• gpt-image-1, which powers ChatGPT’s image generation, is now accessible via the API.
• Developers can control image quality, style, and moderation settings.
• Pricing starts at approximately 2 cents per low-resolution image.
• Leading platforms like Adobe, Figma, and Wix are already integrating this functionality.

This launch follows the remarkable user engagement the feature received within ChatGPT. According to OpenAI, over 130 million users generated more than 700 million images in the first week after its release. This widespread adoption not only underscores the technology's appeal but also highlights the challenges OpenAI faces in meeting demand.

Unlike competing image generators, OpenAI's model stands out for its versatility in creating images across various styles and its ability to accurately render text—a persistent challenge for AI image generators. The API implementation provides developers with granular control over image creation, including options for quality settings (which affect generation speed), the number of generated images, background transparency, and output format selection.

Perhaps most importantly, developers can now adjust moderation sensitivity using the "moderation" parameter, which can be set to "automatic" for standard filtering or "low" for less restrictive filtering. This provides developers with flexibility while maintaining OpenAI's safety measures, such as C2PA metadata watermarking to identify images as AI-generated.

The pricing model follows a token-based structure similar to OpenAI's text models, but with separate pricing for text and image tokens. Text input tokens cost $5 per million, image input tokens cost $10 per million, and image output tokens cost $40 per million. Practically, this translates to approximately $0.02, $0.07, and $0.19 per low, medium, and high-quality square image, respectively.

Major tech companies have quickly embraced the technology. Adobe is integrating the functionality into its Firefly and Express apps, while Figma has already rolled out the model to allow users to generate and edit images from simple prompts. Companies like Airtable, Gamma, HeyGen, Wix, Photoroom, Canva, GoDaddy, HubSpot, and Instacart are either actively implementing or testing the technology.

For instance, Figma users can now utilize gpt-image-1 within Figma Design to generate and edit images, enabling them to adjust styles, add or remove objects, extend backgrounds, and more without leaving their workflow.

This deployment demonstrates OpenAI's ongoing effort to embed its technology across the broader tech ecosystem while maintaining a delicate balance between innovation and safety. As these image generation capabilities become increasingly accessible to developers, we may see a surge in applications across industries—from marketing and e-commerce to education and entertainment—that incorporate sophisticated AI-generated visuals.

Given the explosive growth of image generation within ChatGPT, this API release allows OpenAI to further solidify its role in the developer market while potentially creating new revenue streams through one of its most popular features.