OpenAI’s upgraded image model is now available as API
OpenAI has made its upgraded image generator available to developers through its API, allowing them to integrate the technology into their applications and services. The move comes after the feature was introduced in ChatGPT in late March, generating significant interest and usage.
The new image generator, powered by the “gpt-image-1” AI model, can create images in various styles, follow custom guidelines, and render text. It is a natively multimodal model, capable of leveraging world knowledge to produce realistic images. Developers can generate multiple images at once and control the quality and speed of generation.
OpenAI has implemented safety measures to prevent the generation of content that violates its policies. The model includes safeguards similar to those used in ChatGPT’s image generation feature, and developers can adjust moderation sensitivity to suit their needs. The sensitivity can be set to “auto” for standard filtering or “low” for less restrictive filtering, with the latter limiting fewer categories of potentially age-inappropriate content.
Get your Studio Ghibli filter here: 5 AI tools that still work
All images created using gpt-image-1 are watermarked with C2PA metadata, making them identifiable as AI-generated by supported platforms and applications. The pricing for the service is based on input and output tokens, with costs ranging from 2 cents to 19 cents per generated image, depending on the quality and size of the image.
Several companies, including Adobe, Airtable, Wix, Instacart, GoDaddy, Canva, and Figma, are already utilizing or experimenting with gpt-image-1. For instance, Figma’s design platform now allows users to generate and edit images using the model, while Instacart is testing it for generating images for recipes and shopping lists.