Compare
A · run_95ab4avsB · run_7f5e02code-reviewSwap A and B- Duration
- 8.9s -26%
- A: 12.0s
- Cost
- $0.030 +99%
- A: $0.015
- Tokens
- 12.9k -19%
- A: 15.8k
- Steps
- 7 +17%
- A: 6
- Model changed
- GPT-4.1
- A: Claude Haiku
Steps, aligned
| Change | Step | A | B |
|---|---|---|---|
| Faster | Input policy check | 0.2s | 0.1s |
| Faster | Plan the task | 3.0s | 1.4s |
| Faster | search_codebase | 1.5s | 0.4s |
| Faster | Retrieve context · top 8 | 1.2s | 0.4s |
| Faster | Decide next step | 3.3s | 1.7s |
| Changed | run_tests | 1.9s | 0.3s |
| Added | search_codebase | — | 0.5s |
Output diff
− 36 words + 2 words 0% unchangedReviewed 14 files. Three issues: the cache key ignores the locale, a test is skipped without a reason, and a server action is missing input validation. Tests pass. Suggested changes posted as comments, one marked blocking.(no output)