OpenAI has just developed a new series of AI models that will think more before responding to queries. It has the capability to resolve harder tasks and problems than the previous models. The latest OpenAI series is known as OpenAI o1 and it was previously referred to as the OpenAI’s project Strawberry. With the leadership of Sam Altman, OpenAI unveiled its new model to the market on September 12. There are 2 models included in this series: OpenAI o1 and OpenAI o1-mini. Both these models are designed to emulate human-like reasoning ability by taking more time to analyze before responding.
How does OpenAI work?
The new OpenAI models are similar to human beings and respond slowly to queries. The latest model can resolve PhD level queries including tasks in physics, chemistry, and biology. According to OpenAI, the GPT-4o can solve 13% of these kinds of problems. In the case of o1, it has a success rate of 83%.
OpenAI developed this new model to rule all the main fields including healthcare, physics, and programming. Healthcare researchers can use it for annotating cell sequencing data and programmers can use it for handling multiple workflows efficiently.
The OpenAI o1-mini
The OpenAI o1-mini model is focused on generating accurate information and debugging sophisticated codes. It is mainly developed to help programmers. The new o1-mini is faster and has the capability to provide effective coding solutions to users. Since it is a smaller version of the OpenAI o1, it is 80% cheaper than the OpenAI o1. Budget-oriented users can use it without having any second thoughts.
How to use the latest OpenAI o1 models?
The latest version of OpenAI; OpenAI o1, and OpenAI o1-mini can be used by both ChatGPT Plus and Team users. You can select the required one from the model picker. There is a weekly limit for both models. While the OpenAI o1 model comes with a weekly limit of 30 queries, the OpenAI o1-mini comes with a weekly limit of 50 queries. The company states that they are working to increase the numbers and will allow ChatGPT itself to select the appropriate model according to the level of queries.
The good news is that the company is planning to offer the OpenAI o1-mini model to all the ChatGPT free users.
Safety
The new OpenAI o1 model cannot be bypassed by anyone. These models have passed many jailbreaking tests. While the GPT-4o got 22 scores on a 100 scale, the o1 model got a score of 84. The OpenAI has signed agreements with the U.S. and UK safety institutes to enhance their safety. The authority will continuously monitor the platform to ensure the safety of customers.
Pricing
| Model | Pricing |
| GPT-4o | $5/1M input tokens $15/1M input tokens |
| GPT-4o mini | $0.150/1M input tokens $0.60/1M input tokens |
| OpenAI o1 | $15/1M input tokens $60/1M input tokens |
| OpenAI o1-mini | $3/1M input tokens $12/1M input tokens |
Extra Articles: Apple Watch 10 vs Ultra vs Ultra 2: Which One Should You Choose?
