It’s out – sort of! Yesterday, OpenAI released o1, the first of its “reasoning” models that have been trained to answer more complex questions.
The release of o1, which had been code named “Strawberry”, is a “preview” release and it’s being released alongside o1-mini, a smaller, cheaper version. As noted in OpenAI’s help article on o1, here are some of the things you need to know about the current preview release:
- The o1 series is designed to be helpful for handling complex, problem-solving tasks in domains like research, strategy, coding, math, and science.
- The OpenAI o1-preview and o1-mini models do not have access to the following advanced tools and features: Memory, Custom instructions, Data analysis, File uploads, Web browsing, Discovering and using GPTs, Vision, Voice.
- Conversation topics appropriate for OpenAI’s o1 models include: Strategy ideation, Education and Tutoring, Coding exercises and reviews, Advanced mathematics and physics problems and Complex writing tasks.
- The OpenAI o1-preview and o1-mini models share the same knowledge cut-off as the GPT-4o models, October 2023. However, as I noted here, GPT-4o’s stated knowledge cut-off is misleading, as it can perform browser searches to retrieve information beyond the cut-off date. OpenAI o1-preview and o1-mini don’t have browser capabilities currently, so the knowledge cut-off is presumably fixed.
- Currently, OpenAI o1 models are only available on ChatGPT Paid tiers and for Usage Tier 5 API customers. ChatGPT Enterprise and Edu customers will get access to the OpenAI o1-preview and o1-mini models on September 19. There is no current stated availability date for Free tier customers.
- Users on ChatGPT Plus and Team accounts have access to the 30 messages a week with OpenAI o1-preview and 50 messages a week with OpenAI o1-mini. Better use them carefully!
I didn’t find o1-preview in my drop down of available models in ChatGPT Plus, but I was able to access it here. You can also try it in the API here.
Now that OpenAI released o1, it will be interesting to see what people are able to do with it. Of course, 30 to 50 messages a week don’t provide a lot of room for experimentation, so we’ll see what happens. I’m trying to think about questions to ask it or problems for it to solve that relate to legal and eDiscovery concepts. If any of you have suggestions for those, I welcome them! Once I identify a few test queries to try, I’ll report back with the results.
So, what do you think? Are you excited that OpenAI released o1? Please share any comments you might have or if you’d like to know more about a particular topic.
Image created using GPT-4’s Image Creator Powered by DALL-E, using the term “robot lawyer holding a small strawberry”.
Disclaimer: The views represented herein are exclusively the views of the authors and speakers themselves, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.
