The o3-mini is a leaner model that keeps its pace
US-based AI startup OpenAI has launched a new AI-powered chat feature called’o3-mini’. It said the feature will give users the ability to tap on web searches and call functions from a user’s code. It added that the new model will also incorporate new features, like the ability to tap into web searches and toggle between different reasoning levels.
Crowded airspace has been a problem at Reagan National Airport
US Secretary of State Rex Tillerson has ordered an immediate ban on all flights at Ronald Reagan Washington National Airport (DCA) following the crash of an American Airlines flight and a US Army helicopter on January 30. “This closure will help secure the airspace near DCA,” Tillerson said in a statement. He further said that the restriction would last until the NTSB completed its investigation.
OpenAI has a new reasoning model with a free version
OpenAI is launching a new chat system called o3-mini, which is designed to answer questions in experimentation and science. o3-mini will match o1-mini’s performance in math, coding and science while responding faster. It will be available in both chat and an application. o3-mini should respond 24 % quicker than o1-mini, and provide more accurate answers in the process.
OpenAI and the government collaborated on rival DeepSeek
OpenAI, the company behind ChatGPT and a big part of Stargate, has announced it’s collaborating with US National Laboratories to supply the government with advanced models for things like cybersecurity, energy infrastructure and nuclear security. “The main point for us is that the US has to have the infrastructure to maintain its lead,” OpenAI’s Founder Sam Altman said.
Plane Crash: Everything we know so far
The pilot of the Army helicopter that collided with a civilian aircraft on Wednesday night over the Potomac River in US’ Washington DC, killing all six people onboard, was “fairly experienced”, newly installed Defence Secretary Pete Hegseth said. “They carried out a required annual night evaluation and they did have night vision goggles,” Hegseth added.
The new artificial intelligence arms race is referred to as DeepSeek
DeepSeek has announced that it is testing its new AI-based self-driving car-sharing service, ‘La Llama’. The product will allow users to rent or own a car from a local dealer. It will also let users pay by credit card for purchases, and will be available on the company’s website. DeepSeek is in talks with several major automobile manufacturers to work on the service.
China created an artificial intelligence model that surprised the world
China-based artificial intelligence startup DeepSeek has released an open-source’reasoning’ model called DeepSeek-R1, which can solve some scientific problems at a similar standard to the OpenAI’s LLM. Alibaba released its most advanced LLM so far, Qwen2.5-Max, which the company says outperforms DeepSeek’s V3, another LLM it released in December. Earlier, Moonshot and ByteDance released new reasoning models.
Trump agreed to pay $25 million to settle a lawsuit
Meta has agreed to pay a $25 million settlement to US President Donald Trump over a lawsuit over the suspension of his accounts after the US Capitol insurrection. Trump had sued Meta and Facebook CEO Mark Zuck for suspending him for violating policy against inciting violence and not due to statements made by a member of Congress, as his legal team claimed.
Friends, the man is sleeping at the office of the DOGE
An email sent by Tesla CEO Elon Musk to Twitter employees after Twitter was taken over by him states, “The federal workforce should be comprised of employees who are reliable, loyal, trustworthy, and who strive for excellence in their daily work.” Employees were told to return to work five days a week or resign and receive benefits and pay until September.
There are Ai studies that don’t get much attention
Researchers have developed a new AI model that claims to be the world’s first, capable of stepping-by-step tasks similar to human reasoning at a fraction of the cost and computing power of existing AI models. The model, called ‘DeepSeek-R1’, is capable of step-by-step tasks on data driven scientific tasks taken from real papers in topics including computational chemistry and cognitive neuroscience.