OpenAI has introduced a new subscription tier for its ChatGPT platform, branded ChatGPT Pro, at a steep $200 monthly. This plan is for users needing advanced capabilities in mathematics, coding, and writing. Alongside the subscription, OpenAI rolled out the full version of its o1 reasoning model, which builds upon a preview released in September.
The o1 model employs a self-checking approach to ensure more accurate outputs, reducing common pitfalls associated with generative AI. However, this method also enhances the time required to deliver responses. Compared to its preview version, the upgraded o1 is faster, more accurate, and capable of handling tasks like coding and mathematical problem-solving with improved precision.
“We think the audience for ChatGPT Pro will be the power users of ChatGPT — those who are already pushing the models to the limits of their capabilities on tasks like math, programming, and writing,” Jason Wei, a member of OpenAI’s technical staff, said during a live-streamed press conference on Thursday.
A key addition is the ability to process and reason about image uploads, a feature unavailable in earlier iterations. OpenAI claims that the refined o1 model is designed to be more concise, allowing for quicker responses without sacrificing accuracy. During internal testing, the model demonstrated a 34% reduction in significant errors on challenging real-world questions.
However, there’s a catch: the full version of o1 underperforms in some standardised evaluations. For example, it scored lower than the preview on MLE-Bench, a test designed to evaluate AI performance in machine learning engineering tasks.
While the enhanced reasoning abilities and expanded functionalities may appeal to this audience, the premium price tag raises questions about accessibility and affordability for casual users.