{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifdliuizqigf33novqnkrl2rbsbtd2uidsjvbni2ivagi6etue7qm",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mmyk5o3763o2"
  },
  "path": "/t/improve-performance-on-long-conversations-lazy-loading-memory-management/1382010#post_1",
  "publishedAt": "2026-05-29T12:14:05.000Z",
  "site": "https://community.openai.com",
  "textContent": "**Feature request:** Improve performance on long conversations by using **lazy loading / paginated history loading** together with **message virtualization**. Instead of loading an entire conversation into the browser at once, load only the most recent messages initially and fetch older messages dynamically as the user scrolls upward. For active long-running threads, keep only the visible messages (plus a small buffer) rendered in browser memory, while unloading older off-screen messages and reloading them seamlessly when needed. This would reduce browser memory usage, improve scrolling and typing responsiveness, and keep large project threads fast even after weeks or months of continued use. **As a side benefit, it would also encourage users to stay within existing project threads instead of starting duplicate threads on the same topic just to avoid performance slowdowns, preserving context and improving continuity.**",
  "title": "Improve performance on long conversations (lazy loading + memory management)"
}