Early hands-on says Sonnet 5.5 looks benchmaxxed: pricier and more token-hungry than Sonnet 5

haider1 · x · 2026-09-29

An early hands-on assessment claims Sonnet 5.5 looks "benchmaxxed": scores are strong, but efficiency falls apart — the model is notably more expensive and token-hungry, and even worse than Sonnet 5 on cost and token efficiency in the AA index. The author still prefers Opus 5.5 as the more balanced model, while cautioning it's early days.

Original post →

More from Models

Models channel →