Astra lacks the taste of Opus 5.5, one user says
A user says Opus 5.5 one-shots high-level white-collar work that Astra spins its wheels on, and wastes far fewer tokens.
i’m a little torn because i love how ruthlessly openai pivoted to ensure they had frontier pretraining and i do feel like dots is the right interface and i kinda love openai BUT anthropic always seems one step ahead. they’re not over active on x, they don’t have stores or images or dots. they just have world class models at coding and knowledge work. and taste doesn’t seem to me to be something easy to just hill climb. basically opus 5.5 is oneshotting some high level white collar work that astra spins its wheel on. i have way less dead tokens with opus because it has taste and genuinely understands. and if they’re somehow able to compete on price, it could be. dare i say it. so over.