Benchmarks

What the reviews are

Every review comes from reading each project's source code at a pinned commit. It describes how the project is built: its architecture, how a request flows, what the defaults are, and what is missing or broken. Each claim links to the exact files and lines it is based on.

What the reviews are not

• We did not run any benchmarks. There are no speed, accuracy, cost or quality measurements of the reviewed projects.
• We did not execute the projects to test their behaviour. Statements such as "X retries 5 times" or "Y is off by default" come from the code, not from observed runs.
• Star counts and licences come from GitHub metadata. They are not a measure of quality.

Will this change?

No, not in the near future. The analysis will stay static. This site does not aim to run benchmarks; it aims to offer careful static analysis that you can verify by following the code links.

If you need performance or quality numbers, run the projects on your own workload. Use the reviews to decide which few are worth that effort.