Performance
Read local benchmark results in product terms.
Prefer the live demos? Open this section in the interactive docs.
The /docs#performance section translates bench/baseline-node.json into plain-English speed claims: cheap enough to run while someone types, and fast enough for large imports.
The checked-in baseline uses a deterministic generated-English corpus built from lingo's built-in unit aliases, number forms, qualifiers, ranges, conversions, typos, dates, durations, and sentence templates. caseCount is the number of distinct generated inputs in a suite; samples is the number of timed runs used for the median.
Current local Node v24.5.0 backend baseline:
- Normal single-value field: 492,739 values/s across 900 generated cases, about 2.03 µs per input.
- Mixed natural input: 389,820 parses/s across 791 generated cases, about 2.57 µs per input.
- Strict bulk validation without did-you-mean suggestions: 550,270 checks/s across 240 generated unknown-unit cases, about 1.82 µs per row.
- Sentence scanning: 53,533 scans/s across 260 generated sentence cases.
- 50k-character no-match stress test: 0.341 ms.
Bars show inputs parsed per second (higher is better); µs/op shows the time for one input (lower is better). The caption carries package version, runtime, platform, date, corpus version, and command so the numbers remain a local snapshot rather than a universal promise.