LILYLILYTRUST

Performance Benchmarks

As a new kind of hyperscaler, we have to earn an unusually high bar of credibility. So we hold ourselves to a duty: to regularly put individual parts of our infrastructure through hard test suites and benchmarking, and to publish the results openly, so our customers can see transparently exactly what we subject ourselves to in order to earn their trust.

Current product status
Alpha

Test suites

Every language and framework we support is exercised by its own suite, run against the published platform.

CConformance and correctness across the C standard library and real-world programs.
C
99%
C++OOP, templates, the STL, exceptions and modern C++ correctness.
C++
97%
RustBuild and runtime correctness for Rust crates and binaries.
Rust
95%
GoBuild and runtime correctness for Go modules and services.
Go
98%
PythonRuntime correctness across the Python standard library and popular packages.
Python
96%
JavaScriptRuntime correctness for JavaScript and Node workloads.
JavaScript
94%
Next.jsEnd-to-end build and server-side rendering correctness for Next.js apps.
Next.js
92%

Benchmarks

How our infrastructure holds up under real load, measured and published.

Server Load Test

We test how many apps can run under load on our machines, and how many apps we can fit onto a single machine.

Cold Starts

We benchmark our own cold-start technology, measured against memory usage, start time and more.