☕️ OpenAI's GPT-4.1 Family Debuts as its Newest Flagship AI Coding Model

As the tech industry's programming AI model rivalry reaches an all-time high, OpenAI throws its latest family of AI coders, GPT-4.1, into the heat of battle. Other key highlights include:

• Nvidia unveils plans to bring $500 billion worth of AI chip manufacturing to the US en masse

• Access to OpenAI's future frontier AI models may start requiring government-issued ID verification

• The US' Palantir inks a deal with NATO as it enlists the company's AI-driven Maven Smart System military platform

Join us at AI Tangle as we untangle this week's happenings in AI!

OpenAI releases GPT-4.1 in the API, its latest flagship family of coding models, to take on the competition

After a long period as OpenAI's mainstay and default non-reasoning model, GPT-4o finds itself in the wake of its successor, GPT-4.1, as OpenAI announced the new multimodal family during a Monday livestream. The trio of models - GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano - come with major strides in coding and instruction following, a comparatively much larger context window of 1 million tokens, and an updated knowledge cutoff of June 2024. The models are already available to interested developers through OpenAI's API.

​

However, as the company releases more models, OpenAI is in dire need of GPU capacity.

To free up said necessary GPU capacity, OpenAI will begin slowly phasing out its GPT-4.5 Preview model from the API only after just two months of general availability, along with sunsetting its two-year-old GPT-4 model. As utterly confusing as OpenAI's naming scheme may be, GPT-4.5's reach is being reduced in favor of the GPT-4.1 family due to generally equal or worse performance at ludicrously higher costs. Additionally, with leaks of the company's long-awaited o3 model and even an o4-mini teased here and there, the need for more readily available GPU power seems reasonable.

Nvidia announces plans to bring $500 billion worth of mass AI chip manufacturing to US soil

​Nvidia has recently announced plans to begin producing AI servers in the US for the first time ever over the next four years, looking at up to $500 billion in production value. With a little help from partnerships with TSMC, Foxconn, and Wistron, the move also includes news of production of Nvidia's Blackwell AI chips at TSMC's Phoenix factory in Arizona and a ramp up in supercomputer plants in Texas within 12 to 15 months. Nvidia CEO Jensen Huang said this move "strengthens our supply chains and boosts our resiliency" as the US races to bring back much of manufacturing back home.

​

Palantir secures a contract with NATO to ship its Maven AI military platform to the alliance

​NATO has finalized the acquisition of Palantir's Maven Smart System (MSS NATO), a highly customizable AI military platform designed to improve intelligence fusion, targeting, and operational planning. The MSS NATO is set to be operational within 30 days after a record six-month procurement following the deal's penning on the 25th of March. NATO's acquisition of the system was said to be highly rapid as it is determined to not only boost its defense capabilities but also massively streamline international coordination through the new tech.

​

OpenAI's future models may require organizations to begin presenting government-issued IDs for verification

​According to a support page recently published on OpenAI's website, the company may soon require organizations to go through an ID verification process called Verified Organization to gain access to its top-shelf AI models via its API. The verification process mandates a government-issued ID from supported countries and allows each organization one verification every 90 days. Reasons for the move vary - the system would mitigate unsafe API use, but it would also help to prevent potential IP theft, especially after concerns about data exfiltration by groups potentially linked to DeepSeek.

​

Ilya Sutskever's AI startup, Safe Superintelligence, finalizes a further $2 billion in funding at a $32 billion valuation

Led by OpenAI's co-founder and former chief scientist, Ilya Sutskever's AI startup, Safe Superintelligence (SSI) has raised another $2 billion in funding, according to the Financial Times, putting the company at a $32 billion valuation. Despite having raised $1 billion in September of last year, the company has remained product-less since its inception by Ilya Sutskever, Daniel Gross, and Daniel Levy. However, as the startup's minimalist website states, SSI's one goal and one product remains the same - to develop a safe superintelligence.

​

Former CEO of now-defunct fintech company "nate" has been charged with fraud for faking AI

Albert Saniger, former CEO of e-commerce startup "nate," has been indicted in New York for allegedly defrauding investors by falsely portraying his company's checkout app as AI-driven. Saniger, who raised over $40 million by promising a system with "custom-built deep learning models" capable of processing 10,000 daily transactions, secretly relied on overseas manual labor in the hundreds in a call center in the Philippines. The indictment claims Saniger misled investors by hiding the human labor behind nate's operations, leaving investors with near-total losses following its fallout in 2023.

​Contentful - Create attention-grabbing digital experiences with the help of Conteful's AI-powered toolkit, allowing you to create websites and mobile apps to captivate every audience.

​Gem - Unify your recruiting tech stack with Gem, a platform powered with AI, CRM, and analytics to provide an all-in-one solution to recruiters' problems.

​AlphaRank - AlphaRank is an AI-powered SEO tool developed by Alibaba International's AI Team, offering a personalized, complete solution for businesses looking to improve their search engine rankings.

​Roundtable - Clean your survey responses with Roundtable, an easy-to-integrate API with behavioral tracking, dynamic clustering, and more to cut down on time spent analyzing survey data.

The Journey to GPT-4.5 (46-min watch)

Join OpenAI CEO Sam Altman and a trio of the startup's researchers and scientists as they sit down to discuss the journey of how they got to GPT-4.5 and the knowledge the industry has gained in hindsight.

💼 The Artificially Intelligent Enterprise - Why AI Isn't Matching Expectations​

🔮 AI Lesson - Write Better Prompts with ChatGPT

💡 AI CIO - Owasp Just Sent a Major Signal

🎯 The AI Marketing Advantage - The Age of Forgetful AI Is Coming To An End