Compare
A · run_b70fa9vsB · run_5d8544code-reviewSwap A and B- Duration
- 2.9s -96%
- A: 1m 12s
- Cost
- $0.039 +15%
- A: $0.034
- Tokens
- 16.8k -40%
- A: 28.0k
- Steps
- 7 0%
- A: 7
- Model changed
- GPT-4.1
- A: Claude Haiku
Steps, aligned
| Change | Step | A | B |
|---|---|---|---|
| Faster | Input policy check | 1.1s | 0.0s |
| Faster | Plan the task | 17.7s | 0.7s |
| Faster | search_codebase | 4.2s | 0.3s |
| Faster | Retrieve context · top 8 | 5.4s | 0.3s |
| Faster | Decide next step | 16.5s | 0.6s |
| Faster | run_tests | 5.1s | 0.2s |
| Faster | Write final answer | 18.1s | 0.6s |
Output diff
− 0 words + 0 words 100% 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.