Performance
Benchmarks with an honest boundary
Measured locally in this repository. Reproduce them before publishing comparison claims.
| Operation | pico-intl | i18next | react-intl | vue-i18n | Lingui core |
|---|---|---|---|---|---|
| t() simple lookup | 26.8 ns | 2.90 us | 1.01 us | 40.39 us | 303.4 ns |
| t() interpolation | 572.3 ns | 3.56 us | 1.46 us | 41.02 us | 961.5 ns |
| t() plural | 1.36 us | 6.23 us | 4.48 us | 42.07 us | 7.10 us |
| runtime bundle | 4.2 KB gzip | 13.5 KB gzip | 17.3 KB gzip | 69.2 KB gzip | 2.0 KB gzip |
Read the numbers correctly
- Same-process local benchmark against installed package versions.
- Synthetic workloads cover simple lookup, interpolation, plural/select, and real esbuild browser bundles.
- Do not infer ecosystem superiority from this harness.
- Lingui commonly relies on precompiled catalogs and build-time workflow; interpret its runtime bundle number carefully.
npm run bench:ci
npm run bench:threshold