- OpenAI’s Operator agent will use its logic-heavy o3 model to complete tasks online
- The o3 model will make Operator better at filling out forms and completing purchases as well as less likely to give up when stymied
- The enhanced model may encourage more people to sign up for ChatGPT Pro
OpenAI boasted at the beginning of the year that its Operator was a brilliant AI agent capable of handling everything for you online. If you’re paying $200 for ChatGPT Pro, it could fill out forms, order products, and otherwise use its virtual mouse and keyboard on your behalf.
Now, it’s given the AI agent a brain upgrade by embedding the powerful o3 model in the system, replacing the previous customized GPT-4o model running the service.
OpenAI o3 Operator continues the mission of Operator to interact with the web on your behalf. It can still book a hotel room, order groceries, and even theoretically pay off that parking ticket. But GPT-4o didn’t bring the reasoning power to the service that OpenAI wanted, so the company turned to its o3 “reasoning model.”
The o3 model is supposed to be smarter in a more intellectual way than GPT-4o. It can be more focused and is better at step-by-step thinking. According to OpenAI, employing the o3 model makes Operator “more persistent and accurate when interacting with the browser.” The persistence will help it work through the unexpected obstacles of web browsing, like login requests, pop-ups, and CAPTCHA requests. OpenAI also said that the responses and how the model follows through on long, complex tasks will be noticeably improved.
Operator AI
For most ChatGPT users, this may not matter. Unless you’re ponying up for ChatGPT Pro, Operator is not available. Until then, running a business or managing someone else’s life might make it worthwhile. Otherwise, you might as well stick to the old-fashioned manual approach for menial web tasks.
Delegating them to an AI assistant is an appealing idea, but hardly worth the cost if you’re not going to be using it heavily most days. It would be like hiring a personal chef to butter your toast.
Still, the move to o3 is undeniably a big step forward for OpenAI’s vision of useful AI agents. Of course, OpenAI’s Operator isn’t unique in providing an AI agent at this point. You can find a lot of examples of varying quality, including Browser Use, Proxy 1.0, Hugging Face’s HuggingAgent, and Opera’s Browser Operator.
OpenAI has hinted that it might change the price point for access to Operator, but for now, you have to want to pay ten times as much as ChatGPT Plus costs to get to Operator.
Opening it up to Plus subscribers would be huge in terms of giving access to millions more ChatGPT users, but again, you run into the necessity versus convenience question. Filling in my own forms isn’t worth $200 but if it’s built into the rest of ChatGPT’s features, I might just start letting the AI handle it for me.
Leave a comment