{"version":"2.0.44","anchor":"experimental-reasoning-effort-control","canonical_anchor":"experimental-reasoning-effort-control","heading":"Experimental Reasoning Effort Control","tier":null,"area":null,"scope":null,"heads_up":null,"url":"https:\/\/changelogs.core-directive.com\/v\/2.0.44\/e\/experimental-reasoning-effort-control","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.0.44","markdown":"### Experimental Reasoning Effort Control\n\n**What:** A new system that allows Claude to dynamically adjust how much reasoning it applies to tasks based on an experiment-controlled effort level.\n\n**How it works:** This feature is controlled by the Statsig experiment flag `tengu_effort_exp` with parameter `tengu_effort_level`. When enabled, it accepts three levels:\n\n- `low` (maps to effort value 45)\n- `medium` (maps to effort value 75)\n- `high` (maps to effort value 99)\n\n**Details:**\n\n- The feature automatically injects reasoning effort instructions into Claude's system prompt when the experiment is active\n- Lower effort values prioritize faster, more efficient responses\n- Higher effort values enable maximum reasoning and thoroughness\n- The effort scale ranges from 0-100, with Claude instructed to vary reasoning depth accordingly\n- This is an A\/B test feature - only users enrolled in the Statsig experiment will have access\n- **Evidence**: `function sQI() at line 470523`, `aQI mapping at line 470940`, `system prompt integration at line 470717`\n\n**Technical implementation:**\n\n- New function `sQI()` reads experiment configuration and generates reasoning effort instructions\n- Mapping object `aQI = { low: 45, medium: 75, high: 99 }` converts string levels to numeric values\n- The generated `<reasoning_effort>` XML tag is appended to the system prompt array\n- Experiment values are retrieved via the Statsig client using the `d3()` helper function"}