gpt-6POST
GPT-6 Astra does more than asked, and that makes it slow, one user says
A user says Astra no longer skips tasks but now over-delivers, checking and re-verifying its work, which makes it noticeably slower.
LotusDecoder
@LotusDecoderArchived 2026.09.15Originally posted 2026.09.14
LotusDecoder @LotusDecoderOriginally posted 2026.09.14
以前的老派模型： 叫干嘛，就干嘛，没说不做，说了可能漏做。 现在gpt-6-astra： 做了太多，要查看，要检查，要核对。 大概是没什么风险和问题，终于开始做用户要求的事了。 于是，做事好慢啊。
x.com/lotusdecoder/status/2099307321226846496Archived 2026.09.15 10:12