If you let Claude Code use Opus for small bug fixes, you're paying ~16x more than Haiku for the same result
retained text
Follow-up to my routing post. Plain image this time, lesson learned. Most people leave Claude Code on Opus for everything. I wanted to know what that costs on small, everyday bug fixes. I ran 5 setups on the same real bugs from tomlkit and cachetools, 30 runs total. Hidden tests were added after each run, and I repriced every run from transcript tokens at API list prices. Cost per fix: - Haiku alone: $0.013 (48s) - Haiku + Opus advisor: $0.062 - Sonnet alone: $0.079 (21s, fastest) - Opus + Haiku subagent: $0.126 - Opus alone: $0.206 All 30 runs passed. On a harder single bug (my last post) the gap shrank and Sonnet won. So for small, well-defined fixes, Opus mostly buys you nothing. Caveat: tomlkit might be in training data.
document fetched 2026-10-15 00:43 · canonical https://reddit.com/r/ClaudeCode/comments/1x5waey/if_you_let_claude_code_use_opus_for_small_bug