# YAML benchmarks

YAML performance compared with other .NET libraries on Windows, Linux and macOS: parsing, loading, lossless saving and serialization, with raw results.

Every workload compares OfficeIMO with the other .NET libraries doing the same validated work on one machine. Each run shows all contenders: full and quick runs, on every operating system measured so far. Lower is better.

## No published results yet

The suites for this area are part of the benchmark runner. Results appear here after the next measured run on Windows, Linux and macOS.

## Not measured yet

These comparisons are in the benchmark runner but have no published result. They appear above after their first measured run.

- **YAML object deserialization**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml
- **YAML object deserialization (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml
- **YAML single-value edit and save**OfficeIMO vs YamlDotNet vs SharpYaml vs OneYaml
- **YAML single-value edit and save (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs OneYaml
- **YAML emit from events**OfficeIMO vs YamlDotNet vs SharpYaml
- **YAML emit from events (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml
- **YAML load to model**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml
- **YAML load to model (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml
- **YAML unchanged load and save**OfficeIMO vs YamlDotNet vs SharpYaml vs OneYaml
- **YAML unchanged load and save (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs OneYaml
- **YAML parse to events**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml
- **YAML parse to events (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml
- **YAML object serialization**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml
- **YAML object serialization (.NET 8)**OfficeIMO vs YamlDotNet vs SharpYaml vs VYaml vs OneYaml

[Run the suites yourself →](/docs/capabilities/benchmarks/)

## How to read these results

**Full runs** are complete BenchmarkDotNet measurements. **Quick runs** are shorter diagnostic runs; they are shown so every contender and platform is visible, and carry less weight. "1.6× faster" is the geometric mean of the per-case time ratios, so one large case doesn't dominate. Each run links to its raw JSON with every sample, the source commit and the recorded environment. Measure on hardware close to yours before deciding.

[Benchmark reproduction guide →](/docs/capabilities/benchmarks/) · [Complete Excel matrix and earlier measurements →](/benchmarks/archive/)

<!-- PowerForge markdown artifact sha256:73FCB53CE67C440447A7E00C36F6968ACC9C4DEAAE59394A869EB8B5631D5F31 -->
