insights
ak insights compare
Compare paired benchmark variants
Use ak insights compare for the operation described below. The command help records its arguments, effects, and options.
Command contract
Compare paired benchmark variants
Synopsis
Compare paired benchmark variants. Unknown measurements remain unavailable; invocation success is not quality.
Effects: Read-only comparison; no model calls.
Exit status: 0 success 1 invalid input or unavailable local data 2 invalid arguments
ak insights compare [flags]Examples
ak insights compare --experiment experiment-id --baseline baseline --candidate candidate --json
ak insights compare --helpOptions
--agent string Filter agent definition ID
--baseline string Baseline variant ID
--candidate string Candidate variant ID
--cohort string Filter comparable task cohort
--experiment string Filter experiment ID
-h, --help help for compare
--json Emit machine-readable JSON (implies --no-interactive)
--model string Filter resolved model
--no-interactive Disable interactive prompts (CI-safe)
-q, --quiet Suppress non-error output on stderr
--revision string Filter authored artifact revision digest
--rubric string Select evaluation rubric (default "acceptance")
--runtime string Filter runtime
--since int Inclusive start time in Unix milliseconds
--skill string Filter canonical skill ID
--task-type string Filter task category
--until int Exclusive end time in Unix milliseconds
-V, --verbose Extra diagnostic output on stderr (loses to --quiet)
-y, --yes Assume yes for all promptsSEE ALSO
- ak insights - Measure skill and agent effectiveness