# Multi-Language Code Samples Compared

**Short answer:** Both produce code samples, but a reference generated from the contract derives cURL, Node, Python, and Go from the real operations and responses, so the samples match what the debugger validates. Hand-authored samples are precise and drift. The shared source is the difference.

## Why samples drift

They are written once and copied into the docs. When the API changes, the sample is the last thing updated, and a wrong sample is a direct error for a new developer.

## What generated samples share

| Property | Authored | Generated |
| --- | --- | --- |
| Source | Hand-written | Operations and responses |
| Consistency | As maintained | With the contract |
| On change | Manual | On save |
| Match debugger | Not guaranteed | Yes |

## The failure that matters

A sample that uses a field the API no longer has, or omits a required one, is the classic docs bug. Generated samples reflect the current contract, so they cannot be stale.

## The honest takeaway

Authored samples are the right tool for a small, stable surface with bespoke explanations. Generated samples are the lower-maintenance path for a changing API.

Return to the [Powerduck vs Mintlify comparison](https://www.powerduck.com/blog/powerduck-vs-mintlify/), or read the [OpenAPI guide](https://www.powerduck.com/blog/openapi-guide-hub/).