OpenAI took a bullet for the team, by perhaps scaling the model to something bigger than the 1.6T params GPT4 possibly had and basically telling its competitors

OpenAI took a bullet for the team, by perhaps scaling the model to something bigger than the 1.6T params GPT4 possibly had and basically telling its competitors its not gonna be worth scaling much beyond those number of params in GPT4, without a change in the model architecture