> Employees who tested the new model, code-named Orion, reportedly found that even though its performance exceeds OpenAI’s existing models, there was less improvement than they’d seen in the jump from GPT-3 to GPT-4.
> In other words, the rate of improvement seems to be slowing down. In fact, Orion might not be reliably better than previous models in some areas, such as coding.
The results are what's being reported in The Information article cited, unless you believe that story is false.
A summary of said article (from TechCrunch as the original is paywalled): https://techcrunch.com/2024/11/09/openai-reportedly-developi...
> Employees who tested the new model, code-named Orion, reportedly found that even though its performance exceeds OpenAI’s existing models, there was less improvement than they’d seen in the jump from GPT-3 to GPT-4.
> In other words, the rate of improvement seems to be slowing down. In fact, Orion might not be reliably better than previous models in some areas, such as coding.