One spec, ten implementations

Every Aforo implementation passes the same contract (OpenAPI 2.0.0), seed and conformance suite. This matrix shows the latest result of each one. 6 of 10 implementations have published results.

Runtime numbers are indicative: medians of 5 runs on shared GitHub runners (4 vCPU), which are noisy. Compare orders of magnitude, not single milliseconds. See the methodology.

Latest results

Latest result per implementation, grouped by area. A dash means no data yet.
ImplementationConformanceAPI runtimeArtifactCodeWebFactory
IdStackVerdictDateHurl passSchemathesis failuresRace invariants exactZAP High alertsE2E passCold start (ms)RSS idle (MiB)p95 at 200 RPS (ms)Max RPS with p99 < 250 msImage (MiB)CVE critical / highLOCCoverageInitial JS gzip (KiB)LCP mobile (ms)Cost (USD)Tokens in + outWall clock (min)Human interventions
L01Java + Spring Boot apipending checks2026-11-21100.0 %0yes—————————94.4 %——54.84576,6581300
L02C# + ASP.NET Core apipending checks2026-11-21100.0 %0yes————————————————
L03TypeScript + NestJS apipending————————————————————
L04Go + Gin apipending checks2026-11-21100.0 %0yes—————————85.4 %——40.46409,5991742
L06Python + FastAPI apipending————————————————————
L07PHP + Laravel apipending————————————————————
L08Ruby + Rails apipending checks2026-11-21100.0 %0yes————————————————
L09TypeScript + React webpending checks2026-11-21100.0 %0yes—100.0 %———————93.3 %116.3759————
L10TypeScript + Angular webpending checks2026-11-21100.0 %0yes—100.0 %———————92.9 %124.31,559————
L11TypeScript + Vue webpending————————————————————

Charts

Hurl acceptance pass rateL01: 100.0 %; L02: 100.0 %; L03: no data; L04: 100.0 %; L06: no data; L07: no data; L08: 100.0 %; L09: 100.0 %; L10: 100.0 %; L11: no dataL01100.0 %L02100.0 %L03no dataL04100.0 %L06no dataL07no dataL08100.0 %L09100.0 %L10100.0 %L11no data
Hurl acceptance pass rate. Higher is better; 100 % is required
p95 latency at 200 RPSL01: no data; L02: no data; L03: no data; L04: no data; L06: no data; L07: no data; L08: no dataL01no dataL02no dataL03no dataL04no dataL06no dataL07no dataL08no data
p95 latency at 200 RPS. APIs only; lower is better; indicative
Factory cost to buildL01: 54.84 USD; L02: no data; L03: no data; L04: 40.46 USD; L06: no data; L07: no data; L08: no data; L09: no data; L10: no data; L11: no dataL0154.84 USDL02no dataL03no dataL0440.46 USDL06no dataL07no dataL08no dataL09no dataL10no dataL11no data
Factory cost to build. factory-run.v1; lower is cheaper

Reference oracle

The in-memory oracle in this repository is not one of the ten implementations. It exists only to prove that the suite runs green against a correct implementation.

Latest self-test of the reference oracle (2026-11-21, local Windows 11, Intel Core Ultra 7, 32 GB)
VerdictHurl passSchemathesis failuresRace invariants exactZAP High alertsE2E passNotes
reference: green (ZAP pending)100.0 %0yes——Self-test of the suite; ZAP pending the Docker phase.