OpenAI Unveils New o1 Model with Advanced Reasoning Capabilities

OpenAI has introduced its latest AI model, o1, designed to push the boundaries of artificial intelligence with advanced reasoning abilities. The new model, along with a smaller version called o1-mini, is the first in a series aimed at handling complex questions and tasks with greater accuracy and depth than its predecessors, such as GPT-4o.

Unlike previous models, o1 is built to "think" more deeply before responding, using a "chain of thought" to analyze and refine its answers step by step. This feature allows the model to perform exceptionally well in areas like math, coding, and multi-step problem solving. In tests, o1 significantly outperformed GPT-4o, solving 83% of problems in a qualifying round for the International Mathematics Olympiad, compared to GPT-4o's 13%.

OpenAI Introduces GPT-4o Mini: A Cheaper, More Efficient AI Model

The model boasts an impressive context window of 128,000 tokens and supports up to 16,000 output tokens per request. Its knowledge base is current up to October 2023.

Recorded News — Tech NewsBradley Rhodes

The o1 model has been trained through reinforcement learning, enabling it to recognize and correct mistakes, improving its reasoning skills over time. However, OpenAI admits that the model is slower and more expensive to use, with pricing set at $15 per million input tokens and $60 per million output tokens for developers.

Currently available in preview for ChatGPT Plus and Team users, the o1 model also features improved accuracy, though some challenges like AI "hallucinations"—when models generate incorrect information—remain unsolved. OpenAI has plans to roll out o1-mini to free users soon, but no specific date has been set.