Performance

Both libraries translate the same resources (tests/I18Next.Net.Benchmarks/locales) with the same scenarios. The .NET numbers are the mean of BenchmarkDotNet 0.15.8, the Node.js numbers the median of a batched timing loop after warm-up (tests/I18Next.Net.Benchmarks/node). Measured on an Intel Xeon 2.10 GHz with 4 cores, Ubuntu 24.04.

Scenario	i18next 26.4.2 on Node 22	I18Next.Net on .NET 8	I18Next.Net on .NET 10	Allocated on .NET 10	Faster than i18next
Simple key t("simple")	3,244 ns	221 ns	141 ns	112 B	23×
Nested key t("deep.nested.key")	2,862 ns	218 ns	141 ns	112 B	20×
Interpolation	3,447 ns	447 ns	328 ns	352 B	11×
Three interpolations	4,133 ns	642 ns	447 ns	424 B	9×
Plural	4,888 ns	580 ns	420 ns	536 B	12×
Context and plural	5,100 ns	698 ns	494 ns	656 B	10×
Nesting $t(...)	6,553 ns	1,035 ns	750 ns	1240 B	9×
number format	5,505 ns	819 ns	600 ns	464 B	9×
Fallback language	4,167 ns	462 ns	306 ns	328 B	14×
Other namespace	2,949 ns	249 ns	160 ns	184 B	18×
returnObjects / TObject	17,541 ns	1,551 ns	1,136 ns	2120 B	15×
Missing key	4,664 ns	370 ns	273 ns	352 B	17×

Loaded namespaces are resolved synchronously, the arguments of anonymous objects are read with compiled getters, interpolation runs in a single pass and parsed formats are cached.

dotnet run -c Release --project tests/I18Next.Net.Benchmarks -- --filter '*'

cd tests/I18Next.Net.Benchmarks/node
npm install
npm run bench