OpenAI Launched New Feature: Generate Images In Seconds

OpenAI has upgraded its API with image generation features that let developers create as many images as possible in one attempt. The OpenAI CEO, Sam Altman, made this announcement and shared a concise post stating, “ Imagegen is launched in the openai api! build cool stuff plz.”

Summary:

1. OpenAI has launched an API with image generation features, allowing developers to create as many images as possible in one attempt.

2. The gpt-image-1 model, which follows safety protocols and follows OpenAI’s image generation in ChatGPT, is being tested by major names like Canva, GoDaddy, HubSpot, and Instacart.

3. All photos made with gpt-image-1 are watermarked with C2PA metadata, allowing compatible platforms and apps to identify them as AI-generated images.

Now let’s understand what this upgrade means, how the rollout works with the API to generate images, and how users can effectively utilize this feature.

How OpenAI’s Image Tools Work With API

This update can be integrated directly into your apps or websites based on your requirements. But the interesting part is how much control they would have over it. Basically, you can choose what type of image you want, in what pixels, what background you would prefer, how many images you want, how fast you want them to generate, like anything. You can generate an image based on your expectations or any existing requirements. Interesting, right?

You can also adjust how sensitive the system is when working or checking for inappropriate content, which is helpful if there are any such projects you’re working on.

Although some giant names are already working on it, such as Canva, are finding ways to convert rough designs into a polished one. GoDaddy is doing experiments with the tool in editing and creating logos. HubSpot is also discovering how quickly it can provide marketing images to its users; on the other hand, Instacart is testing the tool for its shopping list designs and recipes.

The model behind this, known as gpt-image-1, has safety safeguards to prevent dangerous content, and OpenAI maintains it does not operate on user information from the API. Pricing is also affordable, with the average photograph costing between 2 and 19 cents, depending on quality.

According to OpenAI, GPT-image-1 follows the same safety protocols as image generation in ChatGPT, including controls that prevent the model from producing content that violates the company’s regulations. Developers can adjust moderation sensitivity, which can be set to “auto” for conventional filtering or “low” for less restricted filtering. According to OpenAI documentation provided to TechCrunch, low filtering minimizes the number of categories of possibly age-inappropriate content.

According to OpenAI, all photos generated with GPT-Image-1 are watermarked with C2PA metadata, allowing compatible platforms and apps to identify them as AI-generated images.

The pricing structure is token-based, like OpenAI’s text models, but with different prices for text and image tokens. Tokens for text input cost $5 per million, image input costs $10 per million, and image output costs $40 per million. In practice, this equates to about $0.02, $0.07, and $0.19 per generated image for low, medium, and high-quality square photos, respectively.

Although this token-based pricing procedure is customizable for developers and available to both startups and enterprise users. OpenAI distinguishes itself by prioritizing responsible AI use. By integrating C2PA information, all AI-generated photos become traceable is a crucial step toward ethical content creation.

The gpt-image-1 API is a significant step forward for OpenAI’s picture capabilities. It helps businesses grow faster using generative pictures while remaining compliant and in control of quality. For developers seeking cutting-edge AI innovation, this is an exciting feature. So, if you’re a developer, try this new feature now to create stunning images.