Review and improve truth
Baseline Comparison now
See what regressed since your last run.
What it does
You point at two runs and see what changed. The scorecard tells you whether the runs are fair to compare, so a regression is real. Ask /evalglass compare for the delta, or baseline to promote a new reference. A standalone visual view is planned.
Say this to your agent
Compare this run to my last one and show me what regressed.
A regression only counts when the two runs are comparable.
Available now — the scorecard says whether two runs are comparable.
In the docs
Read the detail →
what your agent reads for this