The aggressive fringe of the o1 fashions is their potential to excel at reasoning. That is because of the coaching strategy that taught them to assume earlier than responding and, because of this, allowed them to strive totally different methods earlier than touchdown on a solution.
Due to this fact, the o1 mannequin could be very sturdy in math, science, and coding. For instance, it scored 83% on the Worldwide Arithmetic Olympiad (IMO) qualifying examination. For comparability, GPT-4o accurately solved solely 13% of issues. In consequence, when fixing a tough or advanced STEM drawback, this mannequin is your only option.
Regardless that this efficiency may lead you to imagine it’s a higher different in all case situations, that’s not the case. As a result of o1 is an early mannequin, it lacks key ChatGPT options, reminiscent of web searching and accepting media uploads, and because of this, for now, GPT-4o stays the higher choice for many requests.
ChatGPT Plus and Groups customers can toggle between the totally different fashions relying on the immediate, so they do not must decide to utilizing one or the opposite. To make it simpler to resolve which to make use of, OpenAI is engaged on a function that enables ChatGPT to decide on the very best mannequin routinely based mostly on consumer prompts.
