“The initial public reaction to OpenAI's model "5.5" was skepticism, driven by benchmark grids that showed only minor improvements over "5.4".”