Shorely

Benchmarking

PRs run through alternative models — right-sized, over-provisioned, or under-provisioned

Right-Sized

92 PRs

43%

Over-Provisioned

78 PRs

36%

Under-Provisioned

44 PRs

21%

Benchmarking Results by Team

Monthly Savings Opportunity

Based on this month's benchmarks: $10,200 in savings from over-provisioned workloads. $4,600 in quality value from upgrading under-provisioned workloads.

Detailed Results

PRTeamCurrent ModelCurrent CostRecommendedProjected CostQualityStatus
#1247 Add JWT authPlatformOpus$14.20Sonnet$2.80Tests Passover provisioned
#1251 Refactor payment servicePlatformOpus$22.40Opus$22.40Equivalentright sized
#1253 Update landing page copyFrontendOpus$8.60Haiku$0.40Tests Passover provisioned
#1260 ML pipeline optimizationML/AISonnet$6.20Opus$18.80Quality Improvementunder provisioned
#1262 Fix auth token refreshMobileGPT-4o$11.30Sonnet$3.10Tests Passover provisioned
#1248 Add rate limiting middlewarePlatformSonnet$2.80Sonnet$2.80Equivalentright sized
#1254 Component accessibility fixesFrontendHaiku$0.40Haiku$0.40Equivalentright sized
#1261 TensorFlow model exportML/AIOpus$18.80Opus$18.80Equivalentright sized
#1258 Data pipeline schema migrationDataSonnet$4.20Opus$12.10Quality Improvementunder provisioned
#1263 API response cachingMobileSonnet$3.10Sonnet$3.10Equivalentright sized