I measured 681 AI sessions: where your money actually goes
You look at the bill and it makes no sense. You did not feel like you worked more than usual, you asked the same kinds of questions, and the counter doubled anyway. Nobody tells you where it went, so you assume you must be the one asking too much. It is not you. I measured 681 of my own sessions: nine requests out of ten cost almost nothing. What drains your subscription is the moments when the…
You might be surprised to learn where your AI subscription fees are going. A reporter decided to track 681 AI sessions over four months, from April to August 2026, and discovered that only 9 out of every 10 requests cost almost nothing. The real drain on your subscription comes from a select few requests. Specifically, when the AI repeatedly edits the same file without success, it can consume up to four tenths of your entire subscription.
In fact, one such request can cost as much as thirty successful ones. The key to understanding this lies in how the AI behaves during these costly moments. When it keeps hammering away at the same file, it tries multiple times to fix it before moving on. These requests account for only 10% of all requests but consume over half of the subscription cost.
On the other hand, the vast majority of requests (90%) cost almost nothing. The reporter also found that the AI tends to recycle the same solutions over and over once it realizes it can't find a fix. This happens when the AI switches from "let me search" to "let me try something else" without any reason given. At this point, the AI is no longer searching for a solution and is simply permuting through previously tried options.
The reporter learned that re-explaining your project and work to the AI every time is not what costs you money. Instead, it's the AI's tendency to repeatedly edit the same file without success that drains your subscription. To avoid this, the reporter suggests setting an attempt limit for each problem. Two tries should be enough for the AI to search for a solution, after which it should stop and tell you what it tried.
This approach targets the costly requests that consume most of your subscription, allowing you to regain control of your AI usage.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
