Claude Opus 5.5 reportedly arrives with 40% cost cut and faster output

rohanpaul_ai · x · 2026-09-24

Citing investor Gavin Baker on a16z's podcast — frontier labs "always have more advanced checkpoints" — rohanpaulai hints at the next GPT-class model. A quoted thread also lists claimed Claude Opus 5.5 details: Fable 5.1-level performance with typical workload costs cut 40%, input/output pricing of $4/$20 per 1M tokens, cache reads down 60% to $0.20 (all vs Opus 5), output over 30% faster, and a Fast mode up to 2.5x speed at double token prices. Third-party report, unverified.

Original post →

More from Models

Models channel →