Earlier today, OpenAI launched ChatGPT Pro, a $200 monthly subscription for its flagship chatbot. This release is the first of many expected during the next 12 days, as the San Francisco startup has scheduled a slew of announcements to roll out starting today.
Everything from OpenAI’s $20 monthly subscription is included at this price level as well as significantly more access to the GPT-4o and o1 artificial intelligence models. With a ChatGPT Pro subscription—which will cost $2,400 for a full year—users can also use an exclusive model from OpenAI called o1 pro mode that wields more computing power to process answers.
“Power users of ChatGPT, at this point, they really use it a lot, and they want more compute than $20 can buy,” said CEO Sam Altman during the video broadcast announcing the new premium tier. While the hefty price tag may be a shock to many consumers, this subscription is targeted at hyper-engaged users who desire almost unlimited access and at researchers who potentially want to experiment with using ChatGPT for more complex, intensive tasks.
No change to the pricing of OpenAI’s other subscription plans were announced, and the free option remains available. The startup’s first subscription option for its consumer chatbot, called ChatGPT Plus, originally launched in February of last year for $20 a month, and remains that price for now. At the Plus level, users unlock most of ChatGPT’s new features and generative AI models. These subscribers are also not rate-limited as much by OpenAI as free users. How many ChatGPT requests users can make a day, or the amount of time they can spend gabbing with ChatGPT’s best voice interface, is determined by their subscription tier.
The company is targeting its new $200 monthly subscription at those using OpenAI’s generative AI model for more technical work. “People will find o1 pro mode the most useful for hard math, science, or programming problems,” said Jason Wei, an OpenAI research scientist, during the video stream. WIRED has not yet tried a ChatGPT Pro subscription out firsthand to see how it handles these types of requests, though I look forward to testing out the tool as part of helping readers better understand its strengths and limitations, similar to our past work on ChatGPT Plus, as well as its specific features, like Advanced Voice Mode and AI web browsing.
Even though subscribers to ChatGPT Pro receive what OpenAI calls “unlimited access” to the o1 model, GPT-4o model, and Advanced Voice Mode feature, the startup is clear its terms of use still apply. So, actions like sharing an account between multiple people or using the Pro plan to power your own service is not allowed and may get your account banned. Users can request a refund for the $200 subscription within the first two weeks of purchasing it if they are dissatisfied, by going through OpenAI’s online help center.
In addition to ChatGPT Pro, OpenAI announced that the o1 model, which focuses on “reasoning” capabilities and multi-step processing of user input, is no longer in a limited preview. According to the startup, this fully released o1 model answers questions more quickly, can now accept images as inputs, and makes fewer errors. The startup plans to add web browsing and file upload features for ChatGPT’s o1 setting in the future.
As the end of the year approaches, OpenAI is expected to continue launching new AI features. Reporting from The Verge suggests that these year-end releases may include OpenAI’s heavily anticipated generative AI video model, Sora. It’s possible that some of these impending announcements could also provide more insights into how Altman is thinking about AI agents, tools that can potentially perform online tasks on your behalf, and the company’s focus going into 2025.