OpenAI has introduced a new series of reasoning models called “OpenAI o1”, designed to tackle complex problems in fields like science, coding, and math.
What we love especially is the fact that it is positioned as a “Thinking…” model as announced by OpenAI (see the graphics above)
A few key points:
- These models, available from September 12, 2024, focus on thorough reasoning before responding, achieving performance levels comparable to PhD students in challenging tasks.
- The impact of this new model on the LLM field cannot be overstated. By demonstrating human-like reasoning abilities, o1-preview bridges the gap between artificial and human intelligence in ways previously thought to be years away.
- Note currently o1 does not yet access files or images… it will probably come soon.
- It is clear now that for most knowledge worker tasks that are predefined — LLMs are better than ANY human. The key now is moving from task doing, to task defined.
- Is this the end of prompt engineering? today one can ask the model to work with him/her to define the problem and then generate tasks. (agentic AI)
- New highest GPQA score (o1=78.3 vs Claude 3.5S=67.2). Read the full article on GPQA in DigitalRosh.
We warmly recommend checking Dr. Alan Thomson 1 hour video, explaining the OpenAI o1 model on DigitalRosh. For additional info check the Memo, Dr. Thompson’s blog about that.
To sum up, as we always say, you must experience yourself to understand – go for it, and … let’s think together.