Comment by poormathskills

Go look at their past blog posts. OpenAI only ever benchmarks against their own models.