Did this use reasoning or not? GPT-5 with Minimal reasoning does roughly the same as 4o on benchmarks.