Opus 5.5 Cuts Coding Costs for Long AI Sessions

Anthropic's Claude Opus 5.5 is designed to reduce costs for the way developers actually use AI coding tools today. Cached token prices dropped 60%, input and output tokens cost 20% less, and the model runs 30% faster than Opus 5, with overall costs roughly 40% lower for typical workloads. Developer behavior has shifted significantly, with context per request growing 2.6x and Claude working 3.3x longer per prompt. Opus 5.5 addresses this by completing complex tasks in fewer turns, improving cache efficiency, and allowing subagents to inherit parent cache rather than rebuilding context from scratch.