Anthropic Upgrades Claude with Budget Control, Regional Inference, and Advisor Models

Claude Managed Agents gain session budget limits, regional inference options, skill auto-loading, and advisor model integration to enhance enterprise control and execution quality.

Woofun AI reports that Anthropic has updated Claude Managed Agents with four key features to improve controllability and deployment flexibility. The new session budget control allows users to set spending caps, pausing operations when limits are reached until budgets are increased. Enhanced regional inference lets users choose execution locations, with U.S.-only runs costing 1.1 times the standard rate.

Additionally, the system now supports automatic loading of Skill modules from user repositories and introduces an advisor model feature, enabling agents to invoke more powerful models for second-opinion analysis on complex tasks.