I tested Haiku vs. Sonnet across 3 agent tasks – the cheap model won every time

(github.com)

3 points | by aimvik07 4 hours ago ago

1 comments