GPT-6 Pro 明确基于 Astra，Plus 用户却用不到
作者指出 Pro 模式从「同一模型加算力」变成「另一个模型」，Plus 连 Astra 底层都碰不到，是在收回原有权益。
One change here is worth paying attention to: In the past, GPT-X Pro in ChatGPT was essentially just Pro mode. The model underneath was still the same GPT-X used in the regular mode, with the differences mainly coming from the surrounding framework, harness, or other inference setup. Of course, some people speculated that Pro mode might actually use a separately fine tuned model, but overall, it was not an entirely different base model. But now, something has changed. GPT-6 Pro is explicitly described as being based on GPT-6 Astra. It is not based on GPT-6, although there is not even a corresponding GPT-6 model available to users right now. It is not based on the next lower tier model that you can access either. It is based on an entirely different model. And the more important change is this: In the past, different paid tiers could at least have access to the same base model, while Pro mode gave that model more compute, a different harness, or other enhancements. Now Plus users are not even allowed to use the model underneath GPT-6 Pro, which is Astra. Pro mode is effectively turning into Pro model. And that is why I think pushing for Plus users to get Astra in Chat is about much more than saving ourselves one extra click to switch over to Work every time we want to use it. It is about taking back something Plus users used to have. OpenAI is gradually stripping away the benefits of the Plus plan. Remember when they said Plus users would eventually get a chance to experience Pro mode too? Now forget Pro mode. Plus users do not even get access to the regular mode of the base model behind it.
这条是引用着另一条写的：x.com/elainanotes/status/2096158475579756725