{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreif257viq6urj5fy45zhfhyzsdxh6vb242whyi2fkrayxbbmkafnna",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3miizoabb5jj2"
},
"path": "/t/gpt-5-4-ignores-reasoning-effort-none-when-max-completion-tokens-is-used/1378362#post_4",
"publishedAt": "2026-04-02T10:39:33.000Z",
"site": "https://community.openai.com",
"textContent": "Hi!\n\nThanks for the clear repro and raising this!\n\nI can confirm:\n`chat.completions` behaves correctly with `reasoning_effort: \"none\"` on its own, but when adding `max_completion_tokens`, it then ignores `none`, uses the whole budget on reasoning tokens, and returns an empty string with `finish_reason: \"length\"`.\n\nOne possible solution is to switch this call to the Responses API and use:\n\n\n response = client.responses.create(\n model=\"gpt-5.4\",\n input=\"Explain cats to dogs.\",\n reasoning={\"effort\": \"none\"},\n max_output_tokens=100,\n )\n\n\nThat returns normal text with `reasoning_tokens: 0`.",
"title": "Gpt-5.4 ignores reasoning_effort=\"none\" when max_completion_tokens is used"
}