Abstract:
It’s been a few days since OpenAI’s flagship large-scale model GPT-6 Astra was released. At first, the entire Internet was praising its power and shocking, but in the past few days, many people have turned into complaints because of its severe mental decline. Every time a new version of a large model is released, the problem of intelligence reduction will appear within a few days of use. The reasons are also very complicated. It may be insufficient computing power, or it may be deliberately restricted. Judging from the complaints from netizens in domestic and foreign forums,
GPT-6 Astra intelligence reduction has caused the most severe injuries to users in the Chinese area, and many people have even been routed to low-end models
, the performance has seriously declined, and you will be abused even after spending money.
Of course, even if foreign users are not as smart as domestic users, they still suffer from the problem of being unable to connect. In the final analysis, it is still affected by insufficient computing power.
The good news is that OpenAI has also begun to solve the intelligence problem of GPT-6 Astra.
A new version of ChatGPT/Codex 26.908.70816 was launched today
, some netizens reported that they have solved some intellectual problems, but they have only alleviated them and have not completely eliminated them.Whether GPT-6 Astra improves intelligence can be tested using several recognized questions, such as knowledge of date and time, cormorants riding bicycles, etc., but there is now a more reliable test page online, click here.
This page will have three random challenges for GPT-6 Astra to answer. After the results are obtained, they will be pasted back to this page, and then the intelligence reduction situation can be analyzed.
The principle here is related to the training of large models. Each large model is said to generate random numbers, but it often has a specific style. This page relies on this to detect the true identity of the large model.
However, this webpage is only suitable for GPT large model detection
, the results obtained by using other companies' large models to detect have no reference meaning, just like the Agent answers the question of the identity of the large model after limiting the system prompt words.
Comments