Compare runs
B is compared with the previous run of the same agent.
9 steps · 11.0s · ≈$0.0055
9 steps · 11.5s · ≈$0.051
steps 9 → 9 · cost +820% · duration +4% · outcome succeeded
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_b4b56c50 | B · run_d3fc21fc | |
|---|---|---|
| ≠ | #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.