GPT-6 Astra cuts output tokens ~10% per task, sets new Pareto frontier vs GPT-5.6 Sol

ArtificialAnlys · x · 2026-09-04

Artificial Analysis reports that GPT-6 Astra reduces output tokens per task by roughly 10% at max effort versus GPT-5.6 Sol, defining a new Pareto frontier on Intelligence Index vs output tokens per task. This contrasts with the model's 2.5x price hike, which makes it more expensive per task despite the token savings.

Related event: GPT-6 Astra Reviews: Halved Hallucinations but 2.5x Price Hike(18 posts)→

Original post →

More from Models

Models channel →