I’m using Assistant v2. Yes, I’ve updated everthing. But this was also the case in assistant v1. It makes me sick. %95 of the time assistant API gets the information from my files perfectly. But %5 percent of the time, it can’t! And this is unacceptable for professional usage. I’m asking literally the same question. But it says something like: “I couldn’t locate specific information about that.” But it has already given the correct answer to that same question billions of times.
That 5% is so big for business and professional usage. I’m getting tired of not moving myself to LLama 3!