OpenAI's new GPT-5.4 clobbers humans on pro-level work in tests - by 83%

GPT-5.4 is also more reliable, producing 18% fewer errors and 33% fewer false claims than GPT-5.2, according to OpenAI.
The Next Web

OpenAI’s GPT-5.4 sets new records on professional benchmarks

OpenAI released GPT-5.4 today with native computer use, a 1M-token context window, and new professional benchmarks. Find what ...
GamesRadar+

Deltarune Chapter 5 is still underway, and Undertale icon Toby Fox says devs are "polishing" it with "some preliminary internal bugtesting on PC"

When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
CNN

5 things to know for Feb. 6: Nancy Guthrie, TrumpRX, ICE reforms, Boat strikes, Epstein genetic testing

The 2026 Winter Olympics will officially begin today with a spectacular opening ceremony at Milan’s San Siro Stadium. Don’t miss a moment — sign up for CNN’s Milano Memo newsletter to receive key ...
TechCrunch

Meta tests a stand-alone app for its AI-generated ‘Vibes’ videos

Meta is testing a stand-alone Vibes app, the company confirmed to TechCrunch on Thursday. Launched last September, Vibes lets you create and share short-form AI-generated videos and access a dedicated ...