OpenAI last week released the o1 series artificial intelligence (AI) models with advanced reasoning capabilities. These AI models are called the rumored ‘Strawberry’ models that the company has reportedly been developing for the past few months. The company claimed that these large language models (LLMs), consisting of the O1 and O1-mini, were capable of multi-step reasoning and “thinking like a human”. Now, according to a report, these models are being expanded to the AI firm’s enterprise and education customers.
OpenAI o1 AI models are coming to enterprise, education users
As VentureBeat reports, the AI firm is now expanding its latest AI models to ChatGPT Enterprise and ChatGPT Edu customers. This means that users will soon see the option to switch to the o1 model when running a prompt. Specifically, the AI models were released to Plus and Teams users on the day of LLM’s release.
Even though ChatGPT’s paying customers are getting access to the latest AI models, OpenAI has specified that this is still a “in preview” launch, and as such, rate limits will apply. The larger O1 model comes with a rate limit of 50 queries per week. At launch, the rate limit for O1-Mini was also set at the same price, but the company later increased it to 50 queries per day.
The new O1 series models are meant to run and solve complex logic-based tasks, multi-step mathematical problems and areas requiring deep subject matter expertise. The AI firm claims that the AI model will take a few minutes to think about the signal before generating a response. OpenAI highlighted that it mimics the human-like thinking process and allows AI to think about different possibilities and eliminate any errors.
Professor Derya Unutmaz of The Jackson Laboratory, a non-profit biomedical research institute, claimed in a post on Was capable. immunological approach” that would have taken him “days, if not more” to formulate.
Notably, OpenAI has said that the free tier of ChatGPT will also provide access to O1-mini AI models in the coming weeks. It is expected that the rate limit will be less than the amount paid subscribers receive.