Openai on Wednesday brought technology to APIs behind ChatGpt’s new, improved image generation capabilities, allowing developers to integrate it into their apps and services.
Launched for most CHATGPT users in late March, OpenAI’s new image generator went viral with its ability to create realistic Ghibli-style photos and “AI action figures.” It is a complicated blessing for Openai, and it has really strained the company’s capabilities, leading to millions of new sign-ups for ChatGpt. More than 130 million ChatGPT users created over 700 million images in the first week of the tool’s availability, according to the company.
Openai’s API features an AI model called “GPT-Image-1”. A natively multimodal model, GPT-Image-1 allows you to create images across a variety of styles, follow custom guidelines, leverage world knowledge, and render text.
Developers can use GPT-Image-1 to generate multiple images at once and control the production quality, thus controlling the speed.
According to Openai, GPT-Image-1 uses the same safety guardrail as ChatGPT’s image generation. This includes safeguards that restrict the generation of content that the model violates company policies. Developers can control the relaxation sensitivity. This can be set to “auto” for standard filtering or to “low” for less restrictive filtering. Low filtering results in fewer age-appropriate categories of potential content for each OpenAI document provided to TechCrunch.
According to Openai, all images created with GPT-Image-1 are watered out with C2PA metadata, so that they can be identified as AI generated by supported platforms and apps.
The price is $5 per text input token, $10 per image input token, and $40 per image output token. (The token is the fabric of the data the model processes.) According to OpenAI, it converts to about 2 cents, 7 cents, and 19 cents per generated image, respectively, for low, medium and high quality square images.
According to Openai, companies like Adobe, Airtable, Wix, Instacart, GoDaddy, Canva and Figma have already used or experimented with GPT-Image-1. For example, Figma’s Figma design platform allows users to generate and edit images via GPT-Image-1, and Instacart tests models of recipes and shopping list images.