对 OpenAI 当日发布的一手吐槽：Sol、500 美元档与降额
作者点评 GPT-6.1 Sol 接近 Astra 但按 Sol 计价，并批评 500 美元新档、老档额度被砍和 Decision API 定位。
OpenAI dropped a bunch of new stuff today. Here’s my quick take: 1. Dots Some kind of OpenClaw alternative, I guess. I didn’t even understand half of it and honestly I don’t care. 2. GPT-6.1 Sol Good. Actually good. Performance is almost at Astra level, but at Sol pricing. Pretty much exactly what we were waiting for, something that can finally put up at least some fight against Opus 5.5. 3. The new $500 plan with Ultra Fast I mostly don’t care about the plan itself, but I hate the fact that OpenAI were the first to push into the $500 subscription territory. Now Anthropic can easily follow. OpenAI basically paved the road for even more user tier discrimination. And the funniest part is that the new $500 plan gives you 25x Plus limits, while the old $200 plan used to give you 20x Plus. So they nerfed the old plan twice, then came back with 25x for $500. Ultra Fast is also going to burn through your quota insanely fast, so I doubt most people will even need it. The current $200 plan apparently keeps the old limits for another week, then those are gone too. I’m already expecting a pretty big wave of cancellations. Because if $200 is basically 10x now, what’s the point when there’s already a 5x plan? OpenAI don’t deserve $500, and at this point they don’t deserve $200 either. $100, maybe, if you already have Claude Pro at $200. My current setup would be simple: pay Claude $100, or $200 if you need a lot of usage, then throw another $20 at OpenAI because image generation, web chat and the rest are still useful. 4. Decision API Basically a very cheap Jev alternative running on Luna. But Jev is already cheap, so I’m not really sure why I’d care. Especially if this is API-only. Feels like something that should just be included in the subscriptions anyway. There were a bunch of other smaller changes too, but nothing I care enough about to even mention.