As someone who pretty much only uses GPT models (due to how generous the limits are), I always felt like I’d dodged a bullet when people complained about Claude and its load-bearing Claude-isms. GPT models, up to 5.6 Sol, certainly had their own AI-isms, but they always wrote in a way that was easy for me to understand well.
But then Astra came in. I feel like I have to focus ten times as hard to actually understand anything it’s trying to say. The way it writes is just so… dense, as if it’s trying to pack as much information in as few tokens as possible, at the expense of readability.
I wonder if this is due to token-saving optimizations that both OpenAI and Anthropic are doing? Or if it’s just a natural consequence of models getting “smarter”?
Either way, I think I’m switching back to Sol.
Comments URL: [https://news.ycombinator.com/item?id=49797581](https://news.ycombinator.com/item?id=49797581)
Points: 2