OpenAI has announced an improved version of its most capable artificial intelligence model, o3, which replaces its predecessor o1. This new model takes even more time to deliberate over questions, resulting in better answers that require step-by-step logical reasoning. It outperforms o1 on several measures, including complex coding-related skills and advanced math and science competency. Additionally, it is three times better at answering questions posed by ARC-AGI, a benchmark designed to test an AI model’s ability to reason over new problems.
Google, a major competitor of OpenAI, has also revealed its own reasoning model, called Gemini 2.0 Flash Thinking. This shows the fierce competition between the two companies, as they both strive to demonstrate their advancements in AI research. The new models also highlight the trend of AI companies looking beyond simply scaling up models, and instead focusing on problem-solving abilities.
These advancements in AI come at a time when tech giants are making numerous announcements, such as Google’s new version of its flagship model and OpenAI’s various updates, including a free version of its ChatGPT-powered search engine and a way to access ChatGPT over the phone. As the year comes to a close, it is clear that the pace of AI development is only getting faster.
