On my end, Claude's models usually hold up quite nicely vs GPT's. Curious if the compute scaling gap actually shows in model quality or if it's more about availability and throughput.

→ View original post on X — @whats_ai