On Wednesday, the OpenAIA technology was brought back to its new and advanced image generation in its API in its API, allowing developers to integrate their applications and services.
The new image generator of the Openai, which was launched for most ChatGPs in late March, went viral for the ability to create realistic Ghibali-style photos and “AI Action StatisticsThe “It was a mixed blessing for the OpenAI, which leads to several million new signups for ChatzPT, and more than 3 million users have made more than 700 million images in just the first week of the company’s availability.
In the API of Openai, the capacity of the image generation is powered by an AI model called “GPT-Image-1”. A locally multimodal model, GPT -Image -1 can create images in different styles, follow custom guidelines, gain world knowledge and render the text.
Developers can create multiple images at once using GPT-Image-1 and control the quality of the generation-and so on.
According to the OpenAI, the GPT-Emeze-1 ChatGPT hiring the same security maintenance as the image generation, with safguards, with the model that limits a large amount of company’s policy-powered materials. Developers can control restraint sensitivity, which can be set “low” for “auto” or less limited filtering for standard filtering. Low filtering towards the OpenAI document given to TechCrunch limit the low categories of potential age-subjected content.
Opina also says that all the images made with GPT -Image -1 are watercouled with C2 PA Metadeta Thus these can be identified as AI-exposed by supported platforms and applications.
The price for the text is $ 40 per million input token per million input tokens and millions of output tokens per million input token for images. (Tokens are the raw bits of the data that the model processes)) translates about 2 cents, 7 cents and 19 cents for the lower, medium and high-quality square images, respectively.
OpenAI says companies, including Adobe, Airtable, Wox, Instachart, Gaddadi, Canva and Figma, are already using or experimenting with GPT-Image-1. For example, the figure Figma design platform, now lets users create and edit images via GPT -Image -1, while the model is examining the model for the installed recipe and shopping list.
