✨ Introducing OpenAI o1-preview
They have developed a new series of reasoning models for solving hard problems in science, coding, and math.
Explore o1-previewo1-mini spends more time thinking through problems, refining strategies, and recognizing mistakes.
o1-mini excels in generating and debugging complex code, reaching the 89th percentile in Codeforces competitions.
o1-mini performs similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology.
They have developed a new safety training approach that harnesses o1-mini's reasoning capabilities to better adhere to safety and alignment guidelines.
Safety Score: 84/100 (compared to GPT-4o's 22/100)