Monday, April 14, 2025

Start of archive·May 16

Top story

The Agent RaceOpenAI Blog

Introducing GPT-4.1 in the API

OpenAI releases a new flagship model family with measurable improvements in three critical domains (coding, instruction-following, long-context), forcing every company running on GPT-4 or older to reassess their stack immediately.

New model family: GPT-4.1

The briefs

OpenAI's updated Preparedness Framework signals a shift in how frontier AI labs are measuring and mitigating severe harm risks—a critical governance decision that will likely influence regulatory expectations and industry safety standards across the sector.

OpenAI released updated Preparedness Framework for measuring severe harm from frontier AI