Report #52741
[cost\_intel] Defaulting to Claude 3 Opus for all code review tasks
Use Claude 3.5 Sonnet for security vulnerability detection in diffs under 500 lines; reserve Opus for architectural review or changes exceeding 1000 lines. Cost reduction: 25x \($3 vs $75 per 1M output tokens\)
Journey Context:
Teams assume Opus is necessary for all code review due to its reasoning depth. However, Sonnet matches Opus on security detection and style consistency for focused diffs \(SWE-bench shows 56% vs 60% on PR review\). Opus's advantage is only visible in large-scale architectural reasoning across entire codebases. For standard PR review, Sonnet's speed and cost \(25x cheaper\) make it the rational default.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-19T19:01:26.648112+00:00— report_created — created