- OpenAI releases o1, a self-fact-checking AI model
- It excels in complex reasoning tasks
- The model shows promise but has limitations
Meet the new brainiac in town
OpenAI has introduced o1, its latest AI model family. o1-preview and o1-mini are now available in ChatGPT and via API for Plus and Team subscribers.
Enterprise and educational users can access it next week.
Thinking before speaking
o1’s standout feature? It can fact-check itself by considering all aspects of a question before responding.
.jpg)
This “thinking” ability makes it excel at complex tasks requiring multiple steps or synthesizing information from various subtasks.
Impressive, but not perfect
In tests, o1 outperformed previous models in areas like legal analysis and mathematical problem-solving.
However, it’s not without flaws – it can be slower than other models and may still hallucinate occasionally. The tech world eagerly awaits wider access to this promising new AI.
