Thu Sep 19 05:42:01 UTC 2024: ## OpenAI Unveils New AI Models Focused on Reasoning Power

**San Francisco, CA – September 19, 2024** – OpenAI has released a new series of AI models designed to excel in complex reasoning tasks, marking a significant leap in the company’s pursuit of more sophisticated AI capabilities. These new “o1” models demonstrate exceptional ability to solve challenging problems in science, coding, and mathematics, outperforming previous OpenAI models in these areas.

The o1 series consists of two models: o1-preview, an early version, and o1-mini, a faster and more budget-friendly option. While these models shine in reasoning, they are not intended to replace existing GPT-4o models for all tasks. They lack image input support, function calling, and consistently fast response times. Therefore, GPT-4o and GPT-4o mini remain the preferred choices for most developers.

OpenAI has been diligently enhancing the accessibility of these new models. This week, they significantly increased the rate limits for o1-preview and o1-mini APIs for tier 5 developers. Additionally, ChatGPT Plus and Team users have experienced a substantial increase in message limits for the o1-mini model, boosting the weekly quota from 50 to 50 per day.

OpenAI plans to further expand the availability and usage of these models across various developer tiers, and will continue to increase message limits for general consumers as they optimize the models for cost-efficiency.

The o1 series holds immense potential for various applications that demand advanced reasoning, particularly in fields like scientific research, software development, and education. OpenAI’s dedication to refining these models will undoubtedly drive further advancements in the world of artificial intelligence.

Read More