HTTP Status Code

When you visit a webpage, your browser sends a request to a server — and the server always replies with a three-digit number. This number is the HTTP status code. It doesn’t say what’s on the page, but rather how the request went: Did it succeed? Was the page moved? Or is there an error? The most well-known status code is 404, which indicates that a page was not found. But there are well over 60 different codes, each with a clearly defined meaning.

Why status codes are the backbone of the web

The web operates on a principle called HTTP — a set of rules that defines how browsers and servers communicate with each other. Status codes are the formal part of that response. Without them, every app would have to figure out on its own whether a request succeeded — and how.

Status codes are indispensable especially for automated systems. An app that fetches data from another service first checks the status code before doing anything with the response. If it gets a 200, it processes the data. If it gets a 503, which indicates that the server is temporarily overloaded, it tries again later. This mechanism runs entirely automatically — no human needs to intervene.

Search engines also rely on status codes. If a page has moved to a new address and returns a 301 code, the search engine’s crawler knows it should replace the old address in its index. A plain 404, on the other hand, would mean: this page no longer exists, delete the entry.

The system behind the numbers

Status codes are divided into five groups, identifiable by the first digit. The 1xx codes are purely informational, for example: “I have received your request and am still processing it.” The 2xx codes mean success — the most important being 200 (“OK”), which comes back when everything worked. The 3xx codes stand for redirects: the requested resource can be found at a different address.

The 4xx codes indicate errors on the side of the user or their request. The famous 404 belongs here, but so does 403 (“Forbidden”), which means: the page exists, but you’re not allowed to see it. 401, in turn, requires a login. The 5xx codes, finally, indicate errors on the server side. A 500 means that something went wrong on the server itself, without the user having done anything wrong.

The distinction between 4xx and 5xx is important: with a 4xx error, the problem lies with the request; with a 5xx error, it lies with the server. This separation helps developers immediately know where to look for the error.

Status codes in everyday life and in AI systems

As a regular user, you rarely see status codes directly — browsers usually translate them into readable error messages. But in every browser’s developer tools, accessible via the F12 key, they can be observed live for every request. Anyone trying this for the first time quickly discovers that a single page often triggers dozens of individual requests — each with its own status code.

In the AI industry, status codes play a major role because AI models are almost always offered via so-called APIs — interfaces through which programs can query the model. If you use, for example, the API of OpenAI or Google, the server responds to every request with a status code. A 429 means: you’ve made too many requests in a short time, wait a moment. This is especially relevant because AI services often have strict limits on the number of requests per minute.

Even when news reports talk about server outages at major AI services, what’s usually behind it is a wave of 503 errors — the server is reachable, but so overloaded that it rejects requests. Status codes are therefore not just a technical detail, but a direct indicator of a service’s stability and load.