OpenAI has just introduced a new series of advanced reasoning models, known as ‘o1,’ to ChatGPT. Codenamed “Strawberry,” these models mark a significant advancement in AI intelligence. The o1 models excel in science-related topics such as math, physics, chemistry, and coding, and are said to be on par with PhD students.
The exciting news is that two new o1 models – ChatGPT o1-preview and ChatGPT o1-mini – are now available to ChatGPT Plus subscribers. Note that there is a weekly message limit of 30 for o1-preview and 50 for o1-mini. Here’s how you can start testing the new ChatGPT o1 models.
How to Use the ChatGPT o1 Reasoning Model
- Go to chatgpt.com (visit) and sign in with your account.
- Upgrade to the ChatGPT Plus plan, which costs $20 per month.
- After upgrading, click the drop-down menu in the top-left corner and select either ‘o1-preview’ or ‘o1-mini.’
- You can now start interacting with the ChatGPT o1 models. For instance, I tested it with a series of reasoning questions, and it answered all of them flawlessly.
OpenAI’s o1 models excel at handling complex reasoning questions and solving more challenging problems. In my tests, ChatGPT o1 has accurately answered every reasoning question I’ve posed. Previously, even advanced models like GPT-4o and Claude 3.5 Sonnet struggled with common sense reasoning and logical questions.
OpenAI explains that o1 models have been trained to perform chain-of-thought reasoning using reinforcement learning, which is why they may take a moment to ‘think’ before generating a response.
Please note that the o1 models cannot analyze documents, images, or browse the web at this time; they currently support only textual input. OpenAI has indicated that multimodal capabilities will be introduced in the future. For free ChatGPT users, the ChatGPT o1-mini model is expected to be available soon. That’s all for now give OpenAI’s new o1 models a try and let us know your thoughts!
0 Comments