OpenAI o3 vs Google Gemini: Something new is happening every day in the Artificial Intelligence (AI) sector. Two big AI companies – OpenAI and Google are trying to outdo each other. Recently, OpenAI introduced ChatGPT Search to compete with Google search engine, and now it has unveiled OpenAI o3 and OpenAI o3 Mini. The O3 and O3 Mini Reason models were unveiled on the last day of the 12-day 'ShipMas' event.

The new o3 version is an improved version of the existing o1. The o3 model is a family model like the o1. chatgpt The parent company of o3 has also shown a compact version o3 Mini, which can be used for special tasks. OpenAI has given a strong challenge to Google by launching new AI tools.

Focus on reducing risk

OpenAI has indicated that o3 will take Artificial General Intelligence (AGI) one step further in some cases. The company's CEO Sam Altman already wants to prioritize a federal testing framework to monitor and reduce the risks associated with new reasoning models like o3.

Today, we shared evals for an early version of the next model in our o-model reasoning series: OpenAI o3 pic.twitter.com/e4dQWdLbAD

— OpenAI (@OpenAI) December 20, 2024

Researchers will test o3 model

Currently testing of new reasoning models is going on. OpenAI has sought applications from researchers to test these models. The company wants to test these models before completely launching them among the general public, so that there is no scope for error. Researchers can apply for their testing till January 10, 2025.

o3 model performance

It has been claimed in media reports that the O3 model has left behind the performance record of the previous model. This model performed brilliantly in the coding test with 22.8 percent, also known as SWE-Bench Verified.

The o3 model also defeated the Chief Scientist of OpenAI in competition programming. Apart from this, OpenAI's new model has also won one of the toughest math contests, AIME 2024, with a score of 87.7 percent on expert-level science problems.

O3 solved 25.02 percent of the most difficult maths and reasoning problems. It outperformed other models which could not score more than 2 percent. O3 Mini can be launched by the end of January 2025 and O3 can be launched even later.