{
  "id": 2572235,
  "title": "Your LLM App Is Wasting Money: What Happens When Users Close the Tab?",
  "url": "https://urgent.news/2026/08/22/your-llm-app-is-wasting-money-what-happens-when-users-close-the-tab",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-22T12:01:50.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/kristinz/your-llm-app-is-wasting-money-what-happens-when-users-close-the-tab-4k01"
  },
  "original_language": "en",
  "account": "Building an AI chat application involves sending requests to an LLM API. Once a user closes the browser tab, the response disappears. However, the LLM request may still be running if the server doesn't explicitly request cancellation. This can lead to incorrect results and increased costs, as the backend continues paying for unused tokens. In this article, we'll build a cancellation path for a TypeScript LLM server using Hono and Node.js.",
  "summary": "You build an AI chat application. A user sends: \"Explain how distributed systems work.\" Your server calls an LLM API and starts streaming the answer: LLM │ ├── \"Distributed\" ├── \"systems\" ├── \"are\" ├── ... │ ▼ Browser Everything looks great. Then the user closes the browser tab. The response disappears. But what about the LLM request? Is it still running? If your server doesn't explicitly…",
  "key_points": [
    "LLM requests continue after tab closure if server doesn't cancel",
    "Leads to incorrect results and increased costs from unused tokens",
    "Article provides cancellation path for TypeScript LLM server"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}