{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiershfgwh4auclau3nykdoxwjo2ngafqu2j56t5ifi2gsmaz7ra3u",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mmqyhbb4q3z2"
  },
  "path": "/t/codex-ui-extremly-slow-freezing-for-minutes-at-a-time/1381784#post_4",
  "publishedAt": "2026-05-26T11:50:58.000Z",
  "site": "https://community.openai.com",
  "textContent": "I don’t think this is related to hardware or the operating system. There seems to be only one possible explanation: when the context becomes bloated, token calculation may be recalculated multiple times across different tool flows, which I believe causes a freezing issue.\n\nOf course, I can’t say with certainty that “this is exactly how it works and this is why it happens,” because we don’t know precisely how the model operates in the background. However, I don’t think this is a hardware or environment-related issue. I have fairly clear evidence for that: I have the system below, and even I experience freezing issues. On top of that, CPU usage does not increase.\n\nFor now, the solution is this: when you leave it running autonomously, if you have passed 100 turns, you should definitely open a new context window. Because beyond the token count we see in the chat, I believe there is a large data-processing cycle happening in the background on every request, since intermediate expert agents also need to perform calculations.\n\nEven with lightweight requests, intermediate context calculations and tool usage slow the process down.",
  "title": "Codex UI extremly slow, freezing for minutes at a time"
}