{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifc4wr75tfy5fl7hmsmmn63hg6q46vu3tetfyf5h43hoigmvvlq6m",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mkktip33nlk2"
  },
  "path": "/t/the-batch-api-decided-to-make-hundreds-of-requests-that-i-didnt-request-whats-up/1379143#post_8",
  "publishedAt": "2026-04-28T14:49:19.000Z",
  "site": "https://community.openai.com",
  "textContent": "Got another response. They (the bot or the person who’s read bot output to the point that they are picking up some bot-isms) are saying that they were wanting me to get a HAR from replicating the bug. I’m not going to waste my money and time trying to replicate a bug that probably doesn’t even happen all that often.\nThey also state, “API usage is recorded based on processed requests and generated outputs in the system logs. Because of this, charges are tied to what was executed at the system level, even if the expected output file was not returned in the way you anticipated.” Which seems to be in direct conflict with the Batch API FAQ which states “If a batch expires (i.e. it could not be completed within the SLA time window), then remaining work is cancelled and any already completed work is returned. Developers will be charged for any completed work.” Does this not that Developers will only be charged for any completed (and thus returned) work? The logs they are referencing are never “returned” and aren’t accessible through any API (and aren’t even downloadable through the website, weirdly).",
  "title": "The Batch API Decided to Make Hundreds of Requests That I Didn't Request. What's Up?"
}