Anthropic Releases Claude Opus 5.5 with Price Cut and Configurable Thinking

Anthropic has released Claude Opus 5.5, introducing enhanced natural communication capabilities, configurable thinking budgets for reasoning, and lower token pricing. The updated model reduces input costs to $4 per million tokens and output costs to $20 per million tokens compared to Opus 5. As Opus is one of the flagship models with high usage in the AI ecosystem, reducing token costs while boosting capabilities directly lowers operational expenses for developers and enterprises. Additionally, configurable reasoning budgets allow developers to balance latency, token spend, and response accuracy depending on task complexity. Claude Opus 5.5 includes effort controls (ranging from low to max) to set limits on reasoning token generation, though extreme settings can hit the 128,000 output token cap during complex tasks. Prompt cache read costs were also cut significantly from $0.50 to $0.20 per million tokens, facilitating cheaper long-context interactions.