Claude API Supports Dynamic Adjustment of Effort Levels Within a Session
Anthropic has added a beta feature to the Claude API that allows adjusting the `effort` level on a per-message basis. By inserting a system message with empty content and setting `output_config.effort`, you can change the effort level for subsequent responses mid-conversation without modifying the top-level configuration, while still benefiting from prompt caching. The new level takes effect starting from the next user message that includes new input.
Claude API Now Supports Dynamically Adjusting Effort Levels Within a Conversation
Anthropic has introduced a Beta feature for the Claude API that allows you to adjust the effort level on a per-message basis. By inserting a system message with empty content and setting a new output_config.effort, you can change the effort level for subsequent responses mid-conversation without modifying the top-level configuration, while still benefiting from prompt caching. The new level takes effect starting with the next user message that includes new input.
This feature requires the beta header mid-conversation-output-config-2026-07-01 and is supported across multiple models—including Fable 5.2, Opus 5.5, Sonnet 5.5, and Haiku 5.5—as well as platforms like Amazon Bedrock, Google Cloud, and Azure.
Source: Claude Platform Docs
