• by Laurie Sullivan , Yesterday

OpenAI released an image engine for ChatGPT with “thinking capabilities” this week, as it prepares to take the necessary steps to build an advertising platform that can handle more complex queries and images.

"ChatGPT Images 2.0," announced this week, is the next step for the technology -- taking on complex visual tasks and producing precise useable visuals, which are needed for a sophisticated ad platform. The goal is to place and relate objects accurately, the company wrote in a blog post.

“When a thinking model is selected in ChatGPT, Images 2.0 can search the web for real-time information, create multiple distinct images from one prompt, and doublecheck its own output,” the company wrote.

The release of this thinking model is considered a major step toward a professional creative engine. By improving text rendering and allowing for "visual thought" during generation, the platform can now handle the specific requirements of an advertising platform such as rendering accurate user interface (UI) elements, iconography, and marketing layouts.

It can create evolving images and reoccurring storylines that would give an advertising system the ability to pull from the platform.

OpenAI has not provided many details on its advertising platform plans, but it is clear that the focus remains on building a better user experience as the company prepares to go through an initial public offering (IPO).

The strategy, which draws from traditional search advertising services, will give brands more ways to participate, create, and measure ad performance through expanded tools and capabilities based on its ad platform. More to come soon.

OpenAI began testing ads in ChatGPT earlier this year as a CPM model, but rumors suggest that advertisers have been able to test a performance-based ad model with cost-per-click pricing, where brands only pay when a user clicks on an ad rather than paying a flat fee across thousands of views. OpenAI declined to comment.

Open AI did introduce workspace agents in ChatGPT today, described by the company as teams that can handle complex tasks and long-running workflows while operating within permissions and controls set by organization using them.

Google also made key announcements in Las Vegas today around its Gemini Enterprise Agent Platform used to create and manage AI agents and the Agentic Data Cloud that links autonomous agents.