Claude Sonnet 5.5 Burns ~193k Output Tokens Per Task, 7x More Than GPT-6 Astra

ArtificialAnlys · x · 2026-09-29

Artificial Analysis measured that Claude Sonnet 5.5 (max) uses 193k output tokens per Intelligence Index task — the most they have measured and roughly 7x GPT-6 Astra (max) — trading massive reasoning volume for top performance.

At lower effort settings the tradeoff is less favorable: low, medium, and high efforts each sit behind GPT-6 Sol's high, xhigh, and max efforts on the intelligence-versus-output-token curve, meaning Sol achieves higher performance with fewer tokens at comparable budgets.

Related event: Sonnet 5.5 scores 56 on AA Index, 2 points off Opus 5.5—but with record 193k tokens per task(7 posts)→

Original post →

More from Models

Models channel →