@thatlev: Opus 5.5 beat Astra on a coding test while spending about 80% less per task. In Anthropic's FrontierCode v1.1 comparison, Opus at medium effort scores 54.6%, ahead of Astra's best result of 53.3%, at roughly one fifth the cost. It's one benchmark, but the cheaper run still came out ahead. When you're paying a premium for the model, watching the cheaper option get the better result is pretty frustrating #Claude #OpenAI #Opus55 #Astra #Coding

Mat :

Great, now try comparing it with 6.1 Sol ultra

SeducingCamel :

claude ran out of limits shortly after this post

⚛️ :

Currently on my unlimited usage API it feels so good 😭😭🙏🏻🙏🏻

sunbyte :

Pika Cat :

6.1 sol on ultra is way better

Andrea-Leon :

Dan :

Evo_mate :

What's the name of this song?

factorioshmergian :

then gemini argon enters the room

feather :

when

xvrku :

Just use both

xxipbjkj :

WWE

specnox_ :

first to goon to ts

Kitzi :

M.LM_academy :

Variable (★★) :

More crackable design

сын деда твой :

💀