GAX evidence summary — 22 October 2026 retained staging report
Source: headless-commerce/docs/WORDPRESS-STAGING-BENCHMARK-RESULTS.md
Matched 1 CPU / 1 GiB, no swap. Shared host. 640 operations per target, two ABBA cycles.
Selected catalogue/cart mix: native 94.69 ops/s; WordPress 18.89 ops/s; ratio 5.01.
Gax cart operations drove the measured catalogue/cart workflow advantage.
Not proof of whole-platform, equal-cache, end-to-end checkout or conversion superiority.
CMS/PocketBase targets remain open. Ask hello@gax.sh for the full method.
Measured native revision c1e6600c; warm eight-operation guest-cart loop. Historical result, not current-source qualification.
Datasets partially imported and unequal (native/source): 796/802 products, 24621/24736 variations, 262/503 orders. Selected response projections were contract-matched; full dataset equivalence is not claimed.