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

3 points | by aimvik07 6 hours ago

1 comments