The rate limits and cost factor is huge. On my end, Claude's models usually hold up quite nicely vs GPT's for coding quality, but the pricing and limits definitely push people toward Codex for volume work.

→ View original post on X — @whats_ai