Y Combinator's Ankit Gupta says selective benchmarking prevents direct performance comparisons between Anthropic's Opus 4.8 and OpenAI's GPT 5.5

Only one of 15 benchmarks overlapped between the launches