Compare runs
B is compared with the previous run of the same agent.
9 steps · 10.6s · ≈$0.0058
9 steps · 10.8s · ≈$0.046
steps 9 → 9 · cost +689% · duration +2% · outcome failed
What changed between them
| Model | gpt-4.1-mini | claude-sonnet-4-5 |
Step by step
= same (5)≠ changed (3)+ only in B (0)− only in A (0)| A · run_bafce84e | B · run_a1705099 | |
|---|---|---|
| ≠ | #2 gpt-4.1-mini | #2 claude-sonnet-4-5model changed |
| = | #3 sentry.getIssue | #3 sentry.getIssue |
| ≠ | #4 gpt-4.1-mini | #4 claude-sonnet-4-5model changed |
| = | #5 gitlab.searchCode | #5 gitlab.searchCode |
| = | #6 gitlab.readFile | #6 gitlab.readFile |
| ≠ | #7 gpt-4.1-mini | #7 claude-sonnet-4-5model changed |
| = | #8 jira.createIssue | #8 jira.createIssue |
| = | #9 slack.postMessage | #9 slack.postMessage |
System spans are left out; steps are aligned on the tool called (or LLM call), then compared on arguments and outcome.