| ▲ | jychang 2 hours ago | |
What about the juice variable? https://www.reddit.com/r/OpenAI/comments/1qv77lq/chatgpt_low... | ||
| ▲ | tedsanders 2 hours ago | parent | next [-] | |
Yep, we recently sped up default thinking times in ChatGPT, as now documented in the release notes: https://help.openai.com/en/articles/6825453-chatgpt-release-... The intention was purely making the product experience better, based on common feedback from people (including myself) that wait times were too long. Cost was not a goal here. If you still want the higher reliability of longer thinking times, that option is not gone. You can manually select Extended (or Heavy, if you're a Pro user). It's the same as at launch (though we did inadvertently drop it last month and restored it yesterday after Tibor and others pointed it out). | ||
| ▲ | tgrowazay 2 hours ago | parent | prev [-] | |
Isn’t that just how many steps at most a reasoning model should do? | ||