Compare runs
B is compared with the previous run of the same agent.
3 steps · 1.99s · $0.0006
3 steps · 2.76s · $0.0008
steps 3 → 3 · cost +17% · duration +39% · outcome failed
What changed between them
Same app version, agent version, prompt version and model.
Step by step
= same (1)≠ changed (2)+ only in B (0)− only in A (0)| A · run_8a33d8a4 | B · run_b620555f | |
|---|---|---|
| = | #1 route | #1 route |
| ≠ | #2 orders.lookup | #2 orders.lookuporders API timeoutargs, outcome changed |
| ≠ | #3 answer | #3 answercould not answeroutcome changed |
System spans are left out; steps are aligned on the tool called (or LLM call), then compared on arguments and outcome.