Telemetry from ai_reviews
Pass-1 is the default diff review; pass-2 is an optional agentic re-review after pass-1 when enabled.
Final-pass input (code + security)
1.4M
Final-pass output (code + security)
663.1k
Pass-1 input (non-agentic)
984
Pass-1 output (non-agentic)
601.1k
Pass-2 tokens (agentic)
3.2M
Pass-1 share (of pass-1 + pass-2)
15.9%
Avg core IO tokens / PR
24.7k
Avg duration
1m 46s
Total duration (hours)
2.41h
Cache hit rate
71.5%
Est. cost
$57.24
Re-review rate
21.8%
Avg runs / PR
1.53
Activation rate
59.4%
Avg planner iterations
2.3
Avg context retrieved
36.2 KB
Planner token overhead
879.5k
Avg planner duration
13s
Avg cost to non-blocking: $0.73 across 9 resolved PR lifecycles.
Cache hit % is cache-read tokens divided by cache read plus cache-creation tokens, split by whether the review row is the first run on a PR or a later re-review run.
Cohort | Cache hit rate |
|---|---|
| First runsRun index 1 on each pull request | 71.7% |
| Re-runsRun index 2 and above on the same PR | 93.2% |
Activation rate
36.6%
Avg planner iterations
2.2
Avg context retrieved
29.7 KB
Planner token overhead
415.7k
Avg planner duration
12s
Re-runs show cache hits 21.5 percentage points above first runs in this filter.