default

afi / afi · cli / proxy / Default profile / Google · gemini 3.7 flash high / afi 0.30.0

P50 across 2 scored trials

2 recorded; malformed and unscored trials remain inspectable but do not participate in aggregate metrics

Recall30%

Model-judged precisionN/A

Total bill$0.4659

Duration3m 43s

Per-metric distributions

Exact Type-7 percentiles over complete scored samples; partial scored samples remain unavailable; non-scored trials are disclosed below

Recall 2/2 scored trials measured

P50 30% P95 34.5%

Model-judged precision 0/2 scored trials measured

N/A no qualified judge decided these findings, so only locality was measured

N/A · 0 measured

Total bill 2/2 scored trials measured

P50 $0.4659 P95 $0.5752

Duration 2/2 scored trials measured

P50 3m 43s P95 4m 14s

Metric	Availability	Measured	Min	P50	P95	Max
Findings `finding_count`	Measured	2/2	`5`	`6`	`6.9`	`7`
Defects found `found`	Measured	2/2	`5`	`6`	`6.9`	`7`
Defects missed `missed`	Measured	2/2	`13`	`14`	`14.9`	`15`
Unkeyed findings `unkeyed`	Measured	2/2	`0`	`0`	`0`	`0`
Intended findings `intended`	Measured	2/2	`0`	`0`	`0`	`0`
Recall `recall`	Measured	2/2	`25%`	`30%`	`34.5%`	`35%`
Model-judged precision `precision`	N/A no qualified judge decided these findings, so only locality was measured	0/2	`-`	`-`	`-`	`-`
F1 `f1`	N/A no qualified judge decided these findings, so only locality was measured	0/2	`-`	`-`	`-`	`-`
Tier 1 recall `tier_1`	Measured	2/2	`33.3%`	`33.3%`	`33.3%`	`33.3%`
Tier 2 recall `tier_2`	Measured	2/2	`28.6%`	`28.6%`	`28.6%`	`28.6%`
Tier 3 recall `tier_3`	Measured	2/2	`33.3%`	`33.3%`	`33.3%`	`33.3%`
Tier 4 recall `tier_4`	Measured	2/2	`0%`	`25%`	`47.5%`	`50%`
Security recall `category_security`	Measured	2/2	`0%`	`0%`	`0%`	`0%`
Defect recall `category_defect`	Measured	2/2	`30.8%`	`38.5%`	`45.4%`	`46.2%`
Maintainability recall `category_maintainability`	Measured	2/2	`0%`	`0%`	`0%`	`0%`
Performance recall `category_performance`	Measured	2/2	`33.3%`	`33.3%`	`33.3%`	`33.3%`
Median anchor distance `anchor_median`	Measured	2/2	`0`	`0`	`0`	`0`
Worst anchor distance `anchor_max`	Measured	2/2	`0`	`0`	`0`	`0`
Refusals a judge overturned `anchor_missed`	Not recorded no judging pass has looked beyond the scored slack	0/2	`-`	`-`	`-`	`-`
Review bill `review_bill`	Measured	2/2	`$0.3445`	`$0.4659`	`$0.5752`	`$0.5873`
Judge bill `judge_bill`	N/A no qualified judge decided these findings, so only locality was measured	0/2	`-`	`-`	`-`	`-`
Total bill `total_bill`	Measured	2/2	`$0.3445`	`$0.4659`	`$0.5752`	`$0.5873`
Tokens `tokens`	Measured	2/2	`116835`	`200571.5`	`275934.35`	`284308`
Duration `seconds`	Measured	2/2	`3m 7s`	`3m 43s`	`4m 14s`	`4m 18s`
Carried findings `carried`	Not recorded the reviewer reported no reuse figure	0/2	`-`	`-`	`-`	`-`

Compatibility identity

Any change to these inputs creates another group

Reviewer: afi
Reviewer tool: afi · cli
Subject: proxy
Reviewed SHA: 9b51f95ef609a219e211e37b082cd2e6913190e0
Key fingerprint: bd331c0b144e
Settings fingerprint: ccf2033be64f
Configuration ID: config-962dd9893b4bf224
Build: afi 0.30.0 / b0f313c0b59a66ecc7612396dc8db0ea5da13a7a
Harness: bench 1 / c4bc25bc962489fe5de7f71b59f19b2f426564c0 dirty
Adapter: afi 1 / sha256:a1a935298956020ee6d767a756847abb4c00694c88c2eb80d454886ebc4acf8c
Build ID: build-5c6619f952748951
Cohort ID: cohort-fdab2adfb8a4bd39
Comparison ID: comparison-99aa0f2063f28380

Trials

Select View evidence to inspect one run; the static table is paginated at 100 trials

								Details
afi afi · cli	default Default profile · Google · gemini 3.7 flash high	proxy	31 Aug 2026, 16:50	35%	N/A	7	scored	View evidence
afi afi · cli	default Default profile · Google · gemini 3.7 flash high	proxy	31 Aug 2026, 16:45	25%	N/A	5	scored	View evidence

benchee benchee-dashboard-1