Curated
Importance:
Medium
Anthropic launches Claude Opus 5.5: top-model performance at 40% lower cost
Anthropic presented Claude Opus 5.5 on September 22, the first model of its Claude 5.5 family.
What changes for the business is not the capability but the cost per task: the same work with agents, especially work that reuses context, comes out considerably cheaper. If you ruled out a use case on price a few months ago, redo the calculation with the cost per completed task, not per token. And take note of what Anthropic admits: prior tests no longer guarantee behavior.
