Benchmark finds Haiku alone passed every check at 93 percent less cost

What the research found

Benchmark of Claude Code delegation over 28 tasks finds Haiku 5.5 alone passed every check at about 93% lower cost than Sonnet, while outsourcing to external models via an MCP handoff cost more; Opus alone had the best quality.

Claims and how well the evidence supports them

1. Haiku 5.5 alone passed all 84 checks at about 93% lower cost than Sonnet 5.5 alone 96%
2. Handing work to an outside model via the MCP tool cost more than Sonnet alone 90%
3. Opus 5.5 alone scored 99 on blind-graded backlog items and was the only setup with no major fixes 98%
4. The benchmark tasks and raw results are public on GitHub 99%

Evidence

• reddit.com https://reddit.com/r/ClaudeAI/comments/1xclm8k/does_delegating_save_claude_quota_i_measured_6
• github.com https://github.com/gabrielqfr/skrode

Actionability rated medium.